‘Choice architecture’ and socially embedded action – some commonalities

Central to sociological analyses of behaviour, in contrast to predominant accounts within economics and psychology, is the significance of social norms and the manner by which individuals’ choices and actions are influenced by social structures.Footnote ¹ Although one simple position on the structuring and/or agency of individual behaviour proves elusive, recent analytical approaches (e.g. Bourdieu and Wacquant, Reference Bourdieu and Wacquant1992; Archer, Reference Archer2000) note the interdependency between social structures (such as class, gender and ethnicity) and agency over time. At any one moment, human decision-making and actions are constrained by, enabled through, and a reaction towards implicit and explicit assumptions that are a product of the person's history of interactions within particular social contexts or fields.

Behaviour is thus embedded within certain normative frameworks which are provided for the actor by a socio-cultural context, structured through her class, gender and ethnicity. This sociological position would seem quite compatible with earlier understandings of action offered by Sunstein (Reference Sunstein1991: 10), who refers to ‘preferences’ which ‘result from’ certain ‘unjust background conditions’. As Qizilbash (Reference Qizilbash2009: 19) underlines, one important theme within Sunstein's work has been ‘that preferences are always the product of the environment and so to the degree that the environment is a factor which forms preferences, and is “chosen” (if only as a default) by society, [structural] “interference” in people's actual preferences is “inevitable”’.

Such explicit recognition of the influence of a socially structured environment is not apparent within Nudge, yet the central notion of ‘choice architecture’Footnote ² and its inevitability would seem to be influenced by Sunstein's earlier considerations. Indeed Thaler and Sunstein seek to distance themselves from rational choice accounts prevalent within economics and other social sciences by delineating between human judgement and decision-making, and that of Homo economicus or ‘econs’ (2008: 7). Rather than making decisions within a social vacuum, one behavioural tendency receiving emphasis is that of ‘following the herd’; a number of ways in which socio-cultural norms and networks influence human behaviour are cited in substantiating this (ibid.: 59; Ormerod, Reference Ormerod2010).

The sections which follow will outline several senses in which the Nudge perspective might be viewed as problematic from a sociological perspective. Nonetheless the common ground referred to here may be a useful starting point for constructive critical engagement. Recent policy-making (at least in the UK) has become increasingly blind to social structures (Rose, Reference Rose1996) – especially class. The embrace of Nudge by policy-makers may represent a path through which the structures which impact on choices can be emphasised – though first the horizon of ‘choice architecture’ must be broadened in conceptions of time, materiality and ‘the social’.

A narrow architecture around choice – time, materiality and the social

The move away from Homo economicus apparent within Nudge is part of a wider trend within behavioural economics (e.g. Akerlof and Shiller, Reference Akerlof and Shiller2009) and domains of psychology interested in judgement and decision-making (e.g. Hammond, Reference Hammond2007). Amidst the interaction of this latter field with risk research, Slovic (Reference Slovic2000) has argued against the prevailing tendency by which narrow notions of ‘rationality’ restrict conceptions of appropriate agency: ‘The public is not irrational . . . The public is influenced by worldviews, ideologies and values. So are scientists, particularly when they are working at the limits of their experience’ (2000: 411 – cited in Wilkinson, Reference Wilkinson2010: 41).

Unfortunately, a recognition of such wider influences upon lay choices and behaviour, and moreover those of experts, is largely absent in Nudge. Instead the factors which bear on choices are decidedly narrow in terms of time-frame, the material reality which shapes preferences and the variation in preferences across different socio-cultural contexts. Each of these factors – time, materiality and the social – are at least visible within Thaler and Sunstein's analysis, yet the depth of these factors and their corresponding potency are underestimated.

To return to the oft-cited school/work cafeteria example (Thaler and Sunstein, Reference Thaler and Sunstein2008: 1), the placement of fruit in a more visible, attractive location may well have an effect on the numbers of people who choose this. Yet it would be unwise to assume that all people are similarly amenable to being nudged (John et al., Reference John, Smith and Stoker2009). For those whose socio-cultural background means they have a history of eating and appreciating fruit, but who perhaps have fallen into a ‘status quo bias’ (Thaler and Sunstein, Reference Thaler and Sunstein2008: 37) of regularly going for chocolate cake, then the new layout might well nudge them to change their pattern of dessert. Similarly, there may be those who plan on eating fruit, but who often lack sufficient ‘self-control strategies’ (ibid.: 47), for whom a less visible chocolate cake is effectual. In these instances, the factors which bear upon decisions are relatively short-term – momentary lapses of self-control or weekly habits. For those who have seldom eaten fruit over their life-course, and who as a result find it an unfamiliar and/or distasteful alternative, the positioning of fruit is less likely to nudge their selection.

Scenarios such as this raise the question as to whether some people are more ‘nudgeable’ than others – although because Nudge's understandings of time (and the deep rootedness of norms and corresponding behaviour) are limited, this question is lost within the analysis. One ‘solution’ might be to nudge a little more firmly. Making fruit cheaper and chocolate cake more expensive could be seen as a ‘trivial cost’ to induce behaviour change. Material differentiation (through price) may, over a number of years, have contributed to class-based socio-cultural differences in preferences for fruit, where poor neighbourhoods may have less straightforward access to ‘healthy’ produce (Chung and Meyers, Reference Chung and Meyers1999) and instead high-fat, high-sugar food is especially cheap (Cummins and Macintyre, Reference Cummins and Macintyre2002). Yet these norms, rooted in the material and becoming deep-seated over extended periods of time, are not easily eroded. Hence the use of prices as a means of nudging (Thaler and Sunstein, Reference Thaler and Sunstein2008: 106) may correspondingly lack efficacy, acting instead as tax on various socio-cultural norms and tastes which are out of line with those who direct policy.

Considering nudge in terms of its effectiveness: field and habitus as a more nuanced framework for linking architecture and action

One approach which is rich in its considerations of time, materiality and the social, and yet, like nudge, focuses on the linkages between socio-material ‘architectures’ and formats of behaviour, is that set out by Bourdieu (see Bourdieu and Wacquant, Reference Bourdieu and Wacquant1992 for an overview), centring around the concepts of field and habitus. At the heart of Bourdieusian social science is an attempt to develop concepts which are capable of linking the material to the mental through accurate accounts of the social, but where the impact of architectures (‘fields’) on dispositions (‘habitus’) towards behaviour is robustly historical and sensitive to relations of power.

Where Nudge hinges around a conceptual dualism between ‘two systems’ of thinking – ‘one that is intuitive and automatic, and another that is reflective and rational’ (Thaler and Sunstein, Reference Thaler and Sunstein2008: 21) – notions of habitus emphasise action as simultaneously conscious and less-than-conscious. This bridging framework has the potential to enable a much more nuanced approach to considering the extent to which, for example, approaches to and forms of calculative reflection are themselves embedded within dispositions that have been inculcated over many years. In this framework, mental structures are very much the product of the social, as is increasingly recognised within neurological disciplines (e.g. Murray, Reference Murray2008).

The extent to which decision-making adapts to, and is gradually shaped by, environments which are social in nature (Gigerenzer, Reference Gigerenzer2008) draws our attention to the extent to which predisposed tendencies in ‘perception, appreciation, and action’ are very much ‘socially bounded’ (Bourdieu and Wacquant, Reference Bourdieu and Wacquant1992: 126; cf. Simon, Reference Simon1982). Methodologically, this requires us to shift our level of analysis away from individuals and towards understanding the historical and power-related conditioning which shapes decisions. Research pertaining to ‘following the herd’ begins to scratch the surface of this social mechanism, but more depth is required to consider how these socio-cultural processes develop over time to shape behaviour:

Sociological research, in its sensitivity towards varied social contexts, its multi-layered format of analysis and its ‘real world’ research designs, has much to contribute here. The heterogeneity of decision-making contexts, and thus logics, across different communities – to which the above quotation refers – is usefully captured within a Bourdieusian study of risk taking by young people (Crawshaw and Bunton, Reference Crawshaw and Bunton2009). Qualitative research involving young men in a marginalised area in the north of England indicated the pervasiveness of ‘a gender specific form of habitus which determines the practices of young men through requiring them to demonstrate a particular form of tough working class masculinity’ (ibid.: 279). Thus the decisions of these actors towards ‘risky’ behaviour such as violence and drug use, and moreover the extent to which these behaviours were considered as risky when contrasted with the risks of ‘deviating from accepted norms and practices’ (ibid.), are decisively embedded within a particular social context.

A common indifference to these varied socio-historical structures and their influence over decision-making priorities and logics renders the evidence-base on behavioural interventions problematically deficient. While the example cited above is focused upon one marginalised group, it nonetheless draws attention to the factors which influence differing responses to health (and other) risks and the extent to which nudges or other forms of interventions are likely to be of variable effectiveness. The more deeply ingrained a habitus, as inculcated by powerful architectural forces (fields), the less influential more modest forms of ‘choice architecture’ (nudges) are likely to be; though this supposition itself would need to be explored through multi-method research.

As noted already, one of the strengths of nudge, and some associated approaches within certain streams of psychology and behavioural economics, is a more realistic account of human motivations and decision-making which is contrasted against rational-actor ‘econs’. The utility of Bourdieusian and other sociological methods is in many ways an extension of this argument (Bourdieu and Wacquant, Reference Bourdieu and Wacquant1992: 124–5). Further grounding understandings of behaviour in the concrete empirical settings which accurately mirror the real world environments in which policies would be applied, and a recognition of the varied patterns, logics and subjectivities of behaviour that emerge across these social environments, would add much to the quality of the evidence-base.

Considering nudge in terms of its appropriateness: an empirical question

As a format of dispositions bearing upon ‘perceptions and appreciations’, habitus is not only a relevant concept for considering variations in the instrumental effectiveness of nudging across different groups, but also its political appropriateness. Behind Nudge lie profound Enlightenment assumptions where rationality, by its development and application, emancipates society. Sunstein and Thaler (Reference Sunstein and Thaler2003: 1163) contest ‘the false assumption . . . that almost all people, almost all of the time, make choices that are in their best interest or at the very least are better, by their own rights, than the choices that would be made by third parties’. Instead evidence-based health knowledge, harnessed through evidence-based nudging techniques, promotes welfare and wellbeing.

The compatibility of goals between nudgers and nudgees is fundamental to the arguments of Thaler and Sunstein and yet, as apparent in the appropriateness of certain alternative logics for particular social environments (see also Gigerenzer, Reference Gigerenzer2008), the universality of this seemingly ‘neutral’ rationality is open to interrogation and critique (Qizilbash, Reference Qizilbash2009; Anderson, Reference Anderson2010: 371). Hammond (Reference Hammond2007: xiv) expounds a number of limitations of rationality, both as an erroneous model of how actors actually make decisions, but also in terms of its ‘failure as a prescription for human judgement’. By differentiating between humans and ‘econs’, Thaler and Sunstein seek to avoid the former weakness, yet in continuing to delineate between rational and non-rational (economic and non-economic) choices, their attachment to one distinct modernist notion of rationality is value-laden and distances their analyses from lived social experience (Gray, Reference Gray2009b: 14; Gray, Reference Gray and Gray2009a).

The suppositions behind Nudge are similar to those of certain attitudes for managing risk, where there is an intrinsic reference ‘to a generally assumed consensus about the amount of risks to be taken and the directions of societal development to be aimed for’ (Zinn, Reference Zinn and Zinn2008: 27). Were populations socio-culturally homogenous, then this assumption would be cogent, but the variation of knowledge frameworks, material circumstances, day-to-day coping strategies and associated perceptions and appreciations across society makes this stance untenable (Dressler, Reference Dressler1991; Bourdieu and Wacquant, Reference Bourdieu and Wacquant1992).

Various ethical-philosophical postulates of nudge are usefully discussed elsewhere (e.g. Qizilbash, Reference Qizilbash2009; Anderson, Reference Anderson2010), however the variation in tastes/preferences and corresponding logics of ‘choice’ referred to above makes apparent that the political appropriateness of nudge is also an empirical question. Fundamental to Thaler and Sunstein's arguments in favour of nudge are suppositions ‘that what makes a nudge beneficial is not that it steers people in the direction of behaviour which is objectively ‘good’ but rather in the direction of behaviour that they won't regret’ (Anderson, Reference Anderson2010: 371). Given that formats of regret may vary widely, methodological approaches which measure the effectiveness of interventions but which combine this with qualitative insights into participants’ evaluation of being nudged in the light of longer-term experiences and varying goals, and which compare such findings across a wide range of population groups, are necessary.

Such an approach which is sensitive to variation in perceptions, preferences and logics complements understandings of why apparently superior knowledge formats within Western science may fail to resonate with many (cf. Husserl, Reference Husserl1970), as well as strengthening existing arguments that common sense knowledge has its own rationality which is adequately legitimate within its own environment (O'Neill, Reference O'Neill1995; Kay, Reference Kay2010). The limitations of probabilistic knowledge derived from the past, in terms of its continuing relevance for the future more generally, and in its applicability to individual cases more particularly (one never knows if one is the exception to the general ‘rule’, such as dying early from smoking – Alaszewski and Brown, Reference Alaszewski and Brown2007) further affirm this position, underlining the problematic conflation of uncertainty, probabilistic risk generalisations and knowledge in the positivist interpretation (Keynes, Reference Keynes1936; Gray, Reference Gray and Gray2009a).

This line of criticism represents a basis for exploring the varying effectiveness of certain approaches and the legitimation problems that may result, especially when nudge policies are held to be in the interests of marginalised groups whilst applied behind their backs. When employed across different socio-cultural spheres, concepts of field and habitus enable analyses which are sensitive to the range of different goals held by diverse peoples across society as well as the different configurations of priorities that actors may choose between these ends (as recognised within certain corners of behavioural economics). Once again these latter variations may become manifest through different considerations of time, material circumstances and social norms. Certain longer-term health consequences may be balanced against shorter-term strategies for coping with stress (Dressler, Reference Dressler1991), as made available by local norms (e.g. smoking).

Useful models for understanding how shorter-term experiences and longer-term health outcomes are related to one another exist within behavioural economics, such as hyperbolic discounting (e.g. van der Pol and Cairns, Reference van der Pol and Cairns2002). Yet these typically fail to capture the way preferences and logics for acting will vary depending upon socio-economic environment. In contrast, Dressler's (Reference Dressler1991) study of health experiences and behaviours within a marginalised Southern Black community in the United States refers to the ‘profound sense’ of ‘helplessness’ and ‘hopelessness’ which marked participants’ lived experiences, and thus decision-making, as rooted within material deprivation. Corresponding perceptions and appreciations of certain behaviours, alongside expectations around future health status, were markedly different to those of many other socio-cultural milieus. The extent to which the long-term future is considered as something amenable to proactive control or alternatively as fatalistically beyond an individual's capacity to change, as rooted in socio-economic environment (Brown and Vickerstaff, Reference Brown and Vickerstaff2011), will bear upon views of choice architecture and its relevance and appropriateness for different communities of individuals. Such a range of ‘different distinctions’ around goals, priorities and corresponding logics and values (Japp and Kusche, Reference Japp, Kusche and Zinn2008: 79), as facilitated through sociological investigation, require exploration in order to evaluate the appropriateness of interventions.

Wider impacts of nudge: the potential for unforeseen consequences

The arguments presented in the preceding sections include concerns that some people may be more ‘nudgeable’ than others and moreover that the legitimacy of behavioural interventions will vary across different socio-cultural groups. Accordingly, it is possible that the groups within society which policy-makers may most want to nudge (for example towards healthier lifestyles) may be in a number of ways the most resistant to these nudges, with the explanation of this latter issue bound up with the processes which render these groups marginalised and experiencing poorer health outcomes in the first place. That these ‘undesired’ outcomes pertain to individuals’ behaviour regarding future wellbeing, in the midst of incomplete knowledge about outcomes, emphasises the salience of notions of risk and uncertainty and accordingly the potential utility of existing sociological frameworks which are focused upon risk governance and action amidst such intervention. Processes relating to the unequal distribution of risks and the corresponding impact of governance in the midst of this inequality are a central theme within the work of Beck (Reference Beck1992). Meanwhile the socio-culturally rooted variations in perceptions of, and practices around, risks and related political values can be usefully explored following Douglas (Reference Douglas1992, Reference Douglas2006).

Governmentality approaches to health risks meanwhile have questioned the neo-liberal ideology of interventions and the way blame is shifted on to the shoulders of individual citizens, absconding state responsibility for health (Peterson, 1997). Prima facie nudge may be viewed positively in relation to one of these concerns, in its modifying the aetiology of poor health away from the individual and on to the choice architecture around him or her – as is especially clear in Sunstein's earlier work referred to above. Yet the nature of power and control, viewed from the Foucauldian perspective upon which governmentality draws, is irrepressibly and inescapably oriented towards individual actors (O'Malley, Reference O'Malley1992). The problematisation of certain irrational (unhealthy) choices by groups of individuals in need of nudging, and the stigma and implicitly critical lens accompanying this, may thus be less benign than initially evident. The wider and deeper impacts of interventions, alongside the encroachment of the shrinking state, thus also need to be included in evaluations which include the longer-term (Jones et al., Reference Jones, Pykett and Whitehead2011).

Following on from the governmentality focus on a neo-liberal agenda are related concerns over the capacity of nudging to significantly modify behaviour and thus make tangible impacts on health outcomes. Nudge was written with the intention of making governmental paternalism around health behaviours more palatable to liberals but may be correspondingly critiqued (from certain positions) as a political tool in justifying a smaller state (Jones et al., Reference Jones, Pykett and Whitehead2011) and a means by which the government can point to ‘interventions’ which are cheap, maintain choice, but which achieve little – especially amongst the most vulnerable and marginalised (as discussed above).

One further branch within the sociology of risk and uncertainty is that which considers the unexpected and perverse consequences of risk governance – for example how attempts at controlling and modifying behaviour may be ultimately self-defeating (e.g. Gellner, Reference Gellner and Gambetta1988; Bevan and Hood, Reference Bevan and Hood2006) or where interventions to reorient behaviour unwittingly draw on problematic or disputed evidence (public health initiatives around breast-screening have been critiqued along such lines, Solbjør et al., Reference Solbjør, Skolbekken, Sætnan, Hagen and Forsmo2010). Deeper questions exist here over the ‘limits’ of successful intervention regarding choices and behaviour (Hood, Reference Hood2010; Stoker and Moseley, Reference Stoker and Moseley2010). In his warning against overly paternal paternalism, Groopman (Reference Groopman2010) cites a number of ‘best practice’ positions pushed by US Government healthcare experts which have later proved misguided and been revised. A nudge (in its maintaining a freedom of choice) is less dangerous than a shove, however a range of questions remain: Who decides? Which nudges are applied? Over which forms of behaviour? Based on what required level of evidence?

Where nudges do lapse into shoves (Marteau et al., Reference Marteau, Oliver and Ashcroft2009) there exists a danger not only that these shoves may be later found to have been in the wrong direction, but moreover that the very practice of shoving may damage delicate balances of social norms and solidarity within healthcare contexts. The ‘trust−control dialectic’ (Brown, Reference Brown2008) where instrumental attempts at heightening control, in impeding the functioning of norms of trust (on which social order is based) (Gellner, Reference Gellner and Gambetta1988), serve instead to further undermine control may be a relevant framework for considering the longer-term consequences of interventions such as a gradual enfeeblement of the social (Rose, Reference Rose1996). For example, nudging people to donate organs by their automatically opting-in would potentially impact upon the valuable gift-relationship which Titmuss (Reference Titmuss1971: 125) has argued benefits wider society as well as donation services.

In particular it may be instructive to differentiate between those interventions which arguably reinforce norms through making them more widely visible – for example informing people if they are drinking more than most – and those which rely on purposive-rational incentives, such as financial inducements to concord with weight loss regimes (BBC, 2010) or anti-psychotic medication. It is the latter which may detrimentally undermine existing norms, in this latter case voluntary concordance (Szmukler, Reference Szmukler2009). The use of interventions which communicatively engage participants (see John et al., Reference John, Smith and Stoker2009), rather than more instrumentally modify action, may be the most appropriate in the longer-run through transforming formats of habitus.