1 Introduction

Experiments in behavioral economics suggest that the context at the moment of choice influences people’s preferences even when the context should be transparently irrelevant to the choices people make. For instance, it makes a remarkable difference for people’s risk preferences whether the information is framed in terms of losses or gains (Kahneman & Tversky, 1986), for moral preferences whether the context makes people evaluate choice options separately or jointly (Bazerman et al., 2011), or for their willingness-to-pay whether people were primed to think beforehand about high or low anchors (Ariely et al., 2003). Moreover, behavioral economists argue that preferences are not only endogenous to situational framing but also to the wider socio-cultural environment (Bowles, 1998; Fehr & Hoff, 2011). In other words, it is not only ‘irrelevant’ features of the decision situation (the choice frame) that induces changes in individual preferences, but the social context individuals are situated in has also a deep and durable impact on their preference formation processes. For instance, exposure to local female leaders affects villagers’ attitudes toward gender roles (Beaman et al., 2009), exposure to soap operas with strong, independent female characters leads to a decline in fertility rates (La Ferrara et al., 2012), or exposure to job opportunities in which women’s autonomy and education is stressed leads to changes in parental preferences (Jensen, 2012). These studies are examples for the evidence behavioral economists have gathered that illustrates that “prolonged (and sometimes even brief) exposure to a given social context shapes who people are.” (Hoff & Sitglitz, 2016, p. 26).

A core idea to explain preference endogeneity are cultural mental models (Hoff & Stiglitz, 2016). Mental models are concepts, categories, narratives, and ideologies that “shape the way we attend to, interpret, remember, and respond emotionally to the information we encounter and possess” (DiMaggio, 1997, p. 274). Importantly, those mental models affect people’s preference formation since they are cognitive devices that help people categorize the world around them, including how people rank products, people, political parties, lifestyles, etc. At any moment in time, people hold multiple mental models that they can draw upon to interpret the situation and the selection of a model “is guided by cultural cues available in the environment” (DiMaggio, 1997, p. 275). In this sense, the context at the moment of choice and the wider social context constantly interact: the wider social context creates the set of mental models people are exposed to over time and situational circumstances prime the use of a specific mental model at a given moment in time.

There are important implications of preference endogeneity for the explanatory power of models in positive economics, e.g., predicting the stability of institutional regimes or the persistence of dysfunctional preference-norm equilibria (Hoff & Sitglitz, 2016; Dold & Lewis, 2022). Yet this paper focuses on implications of preference endogeneity for normative economics, i.e., for ways economists evaluate individual welfare and the relative merits of different social states. If preferences are endogenous, traditional welfare economics stands on shaky methodological grounds. It is not any more self-evident to take preference satisfaction and (potential) Pareto improvements as the relevant normative yardsticks for welfare analysis since institutions and policies are not just regulative and affect behavior via relative price changes, but are also constitutive and affect people’s processes of preference and belief formation (Hargreaves Heap, 2020; Delmotte & Dold, 2021). In light of endogenous preferences, Bowles and Gintis (1993, p. 100) rightly conclude that “it is hard to avoid the issue of the origin of the wants in question, and why these wants, as opposed to others, which could as well have emerged from different initial conditions, should be satisfied.”

In recent years, there has been a lively debate in behavioral normative economics about how to best respond to the challenge of endogenous preferences (Bernheim, 2016; Dold & Schubert, 2018). Broadly speaking, there are three approaches. Approach one tries to substitute an alternative welfare criterion (such as happiness, or subjective well-being) for normative economics’ reliance on preference satisfaction (Layard, 2006; Frey & Stutzer, 2010). Approach two aims to uphold traditional economics’ commitment to preference satisfaction but argues that some form of preference purification is required to enable the economist-analyst to identify people’s ‘latent’ or ‘true’, preferences that fulfil the axioms of neoclassical rationality (Bernheim & Rangel, 2007; Beshears et al., 2008). Approach three aims to overcome the reliance on preference satisfaction altogether and aims to ground normative economics in freedom of choice or opportunities rather than preferences (Sugden, 2018, 2020). Opportunity-based approaches criticize the absence of psychological explanations for the existence of ‘true’ preferences and question the necessary connection between neoclassical rationality (in particular, consistency) and welfare.

All three approaches have advantages and shortcomings. A weakness of approach one—the happiness approach—is that it takes at face value what people report in these surveys when they are known to be subject to psychological biases, e.g., influenced by high or low points of an experience and the most recent feelings associated with an experience. Moreover, the happiness approach is based on the implicit presumption that people want to be happy above anything else. However, people might plausibly have other projects in life (Hargreaves Heap, 2013, p. 993). A weakness of approach two—the preference purification approach—is that it is unclear how to identify people’s ‘true’ preferences (epistemic problem) and, taking the insights on evolving preferences seriously, it is unclear why a person’s core of ‘true’, welfare-relevant preferences should indeed exist as a set of well-integrated, context-dependent preferences (ontological problem) (Dold, 2018). Finally, approach three—the opportunity-based approach—circumvents the aforementioned problems, however it is unclear why opportunity, i.e., a larger choice set, should trump everything else considering the psychological literature on choice overload and people’s interest in self-commitment strategies (Dold & Rizzo, 2021).

Given these respective weaknesses, this paper discusses a fourth response to endogenous preferences advocated by Viktor Vanberg and Carl Christian von Weizsäcker, two prominent proponents of contemporary ordoliberalism. The structure of the paper is as follows: Sect. 2 discusses Vanberg’s approach of constitutional political economy and highlights its potential shortcomings in dealing with the challenge of preference endogeneity. If preferences are context-dependent, the paper argues that voluntary and informed choices cannot be taken as uncontroversial starting points when judging the relative merits of markets and political institutions. Section 3 provides a discussion of von Weizsäcker’s approach to preference endogeneity that overcomes some of the shortcomings of Vanberg’s approach. A core idea of his approach is that normative economics must think carefully about the institutional prerequisites of individual autonomy. However, Sect. 4 argues that von Weizsäcker’s approach still focuses too much on external structural concerns and outcome freedom and neglects concerns for internal agentic capabilities that allow people to navigate large opportunity sets. The section follows Sen (2004) who has argued that the process through which choices are made (process freedom) is something individuals care about beyond the availability of choice options (opportunity freedom). Section 5 discusses policy implications of a joint concern for the process and opportunity aspects of freedom.

2 Contemporary ordoliberalism’s response: Viktor Vanberg’s choice individualism

The political economy tradition of ordoliberalism originated in the interwar period in German-speaking academia.Footnote 1 Ordoliberal scholars study and emphasize the crucial role of both formal and informal rules to foster competition and prevent power concentrations in markets and politics. Over the last four decades, Viktor Vanberg’s research program of constitutional political economy (CPE) has greatly enriched ordoliberal thinking (Dold, 2022). Vanberg has put ordoliberalism on a firmer methodological and normative foundation by connecting it to Hayek’s evolutionism and Buchanan’s contractarianism. Like Buchanan, Vanberg argues that the best way of improving political and economic processes is to reason and reform the legal-institutional framework (the ‘rules of the game’) of both markets and politics.

According to Vanberg (2014), CPE’s normative benchmark for the evaluation of markets and politics is individual sovereignty. Aiming for consumer sovereignty in the economic realm means that markets are to be judged in terms of their responsiveness to consumers’ evolving preferences; and aiming for citizen sovereignty in the political realm means that institutions are to be judged according to their responsiveness to citizens’ evolving preferences. Politics embodies citizen sovereignty by embracing majority rule, regular elections, and other institutions of democracy. Markets embody consumer sovereignty with a system of private law and competition. Despite differences in the respective institutional embodiments, the key takeaway from an ordoliberal perspective is that both markets and politics should be built on the same ultimate source of legitimacy: voluntary agreement among autonomous individuals.

2.1 Choice individualism versus utility individualism

Vanberg explicitly acknowledges the challenge of endogenous preferences for economists’ attempt to assess the relative merits of alternative social states:

The fact that the individual preferences on which the welfare theoretical calculus is based are themselves subject to evolutionary change, [raises] the question of how welfare economics can serve to evaluate policies if the very measuring rod it uses is itself subject to change over time. (2014, p. 329).

According to Vanberg, a crucial distinction in this regard is that between utility individualism, which is the normative foundation of welfare economics, and choice individualism, which is the normative foundation of CPE. Importantly, both strands of individualism are forms of normative individualism that take individual preferences as the starting point for assessments of social states. Yet, Vanberg (2014, p. 330) argues that the two strands differ critically in how they specify that starting point. Utility individualism substitutes the notion that ‘individual preferences are to count’ by the notion that ‘individual utilities are to count.’ Choice individualism, on the other hand, translates ‘individual preferences are to count’ into ‘individual choices are to count.’ This subtle difference has important implications for normative-evaluative analyses in economics.

According to Vanberg, utility individualism is a framework in which—once measured—individual utilities are computed by the economist without taking the individual person as a choosing agent seriously. It defines utilities as that what a person really ‘prefers,’ but not necessarily what she chooses. This understanding of normative individualism is illustrated by recent attempts in behavioral welfare economics to define utility as the satisfaction of a person’s ‘true’ preferences (Sugden, 2018, pp. 53–82). Those ‘true’ preferences are assumed to fulfill the axioms of neoclassical rationality (in particular, consistency); they are counterfactual in the sense that they are what a person would choose had they “complete information, unlimited cognitive abilities, and no lack of self-control.“ (Sunstein & Thaler, 2003, p. 1162). Behavioral welfare economists have used different measuring techniques of how to identify those ‘true’ preferences, among them are structural estimation, active decisions, asymptotic choice, aggregated revealed preferences, reported preferences, and informed preferences (Beshears et al., 2008). Techniques such as these illustrate that utility individualism is ultimately not interested in the person as a sovereign chooser; it conceptualizes an individual agent only as ‘reading or measuring points’ from which utility data are collected in order to derive measures of individual and social welfare on which policy recommendations can be based (Vanberg, 2014, p. 333).

In contrast, choice individualism takes individual choices as the starting point from which normative criteria for judging social matters are to be derived. The crucial point is that preferences are not imagined separately from a person’s choices. Consequently,

Choice individualism regards individuals as the ultimate sovereigns in social affairs whose actual choices … are the only source from which the observing economist can derive conclusions about ‘social welfare’ or, more precisely, the normative status of social transactions, social arrangements, and policy measures. (Vanberg, 2014, p. 331).

According to Vanberg, evolving preferences are not an obvious problem in the CPE framework since the emphasis is on choice (and thus momentary preferences) and not on a notion of utility that assumes rational and time-consistent preferences. CPE argues that “sovereign individuals can base their choices on no other preferences than their present preferences.” (Vanberg, 2014, p. 348). Since the emphasis is on choice and not on utility functions, the idea of rationality is not a normative precondition for welfare-improving choices. Yet, there is a precondition in CPE too, albeit a ‘weaker’ one: for choices to count as normative input into the constitutional political economist’s analysis, actual choices need to be “voluntary and informed” (Vanberg, 2014, p. 331).

Vanberg derives his emphasis on voluntary choice from the idea of mutually beneficial market transactions and extends it to cooperation in social-political arrangements. He (2005, p. 28) states that: “In the case of market exchange the economist’s claim that both sides gain in terms of their own judgement can, in the final analysis, be based on no other evidence than the voluntary agreement of the parties to the transaction.” What does the emphasis on informed choice mean? Vanberg (2014, p. 346) gives the following answer:

Respecting in individuals as sovereigns does not require us to subscribe to the claim that individuals always ‘know best what is good for them.’ There is ample evidence that individuals make what in retrospect they regret as foolish choices, choices that more prudent, more experienced, and more knowledgeable advisors might tell them not to make. It is unquestionable that such advisors—including welfare economists and (libertarian) paternalists—may serve individuals well by providing information that can help them to make more ‘educated’ choices.

Vanberg argues that providing information does not mean to override individual sovereignty: the fact that experts might know better in some instances than the individuals themselves ‘what is good for them’ does not give them authority to disrespect individual choices. CPE accepts that preferences are liable to change and individuals might suffer from limited knowledge and cognitive abilities, yet those ‘shortcomings’.

Can only be reasons for providing sovereign individuals with arguments that enable them to have a more realistic understanding of their likely future preferences and, in general, to be guided by less myopic and better informed preferences. (2014, pp. 347-8).

In a nutshell, the choice-individualist perspective identifies information provision as the main task of normative political economists in the context of people’s evolving preferences.

2.2 A response to CPE’s focus on information as the prerequisite for voluntary choice

But is information provision sufficient to guarantee choices that are autonomous, i.e., choices individuals can fully endorse as their own? Taking the idea of endogenous preferences seriously, this can be doubted. First, there is no neutral way of providing information. In other words, information is part of the choice context and influences the decision-making process in subtle ways the decision-maker is often unaware of (van Emmerick & Dold, 2023). Also, information provision may not be sufficient to circumvent the preference-shaping power of framing and defaults as Halpern et al. (2013) illustrate. In their study, patients with terminal illnesses make choices about end-of-life care treatments. Patients are given two options, (A) comfort care or (B) life extension. Intervention (B) involves therapies that are intrusive and unpleasant; the time added to one’s life is accompanied by being put on a ventilator or having a feeding tube inserted. Intervention (A) circumvents invasive interventions and focuses on comfort by means of systematic pain management. The patient pool is divided into three groups that each receives a different advanced directive. For 1/3 of the patients, option (A) is the default (prechecked box); for another 1/3 of patients, neither (A) nor (B) is preselected; and for the last group, Option (B) is prechecked. The study finds that in the first group comfort care is selected by 77% of subjects, in the second group by 66%, and in the third group by 43%. After the choices are made, the researchers reveal the results to all patients, explain that they have been randomly assigned to three different defaults, and then offer them to change their minds. If information provision is all that is needed to autonomously express one’s preferences, this would be the moment. However, the study finds that out of 132 subjects only two changed their choices. This striking finding demonstrates the stickiness of the choice frame: even when the patients are told what the defaults were, that the defaults have been determined randomly, and the way defaults influence choice, the context-effect persists. Making sure that people have a choice is laudable, and giving them information is even better. However, if people have to make tough choices in complex situations, they are likely to be heavily influenced by the framing of the choice options and take the default despite having been given ample information—particularly in cases where they do not have strong antecedent preferences and lack prior experience.Footnote 2

CPE assumes that people by and large know how to make the ‘right’ choice within a given choice environment if one only provides them with sufficient information. Yet, as this study illustrates, assuming patient sovereignty and ignoring the preference-shaping influence of the context can have a significant impact on autonomy, i.e., the sense of being the author of one’s own life. Taking autonomy concerns seriously, economists cannot simply assume individual sovereignty after information provision (more on this point in Sect. 4). Instead, economist may need to open up their discipline to a more systematic discussion of prerequisites of individual autonomy and acknowledge that autonomy can be more or less present in decision situations, depending on, inter alia, the preference-shaping power of context factors.

In focusing on information provision as a necessary ingredient for autonomous choice, CPE downplays other social-structural influences on individuals’ preference formation processes. For instance, one can imagine in the context of the Halpern et al. (2013) study that patients are influenced by the cultural mental model that they were exposed to before making a choice. It is conceivable that patients, who live in social milieus where serious illnesses are primarily classified as a burden for others, choose differently from patients, who are exposed to milieus where mutual care is seen as a core attribute of the ‘good life’. In other words, while CPE acknowledges that people’s preferences evolve over time, it does not fully acknowledge the intricacies of preference endogeneity for normative analysis, neither in the form of the preference shaping power of the context nor the wider socio-cultural environment. The next section discusses von Weizsäcker’s approach that rectifies some of CPE’s shortcomings in regarding competition among cultural mental models as a core condition for autonomous choice.

3 Von Weizsäcker’s contribution to ordoliberal thinking about preference endogeneity

Von Weizsäcker raises the question of what “the welfare implications” of “endogenously changing tastes” are while emphasizing that in “a world with changing tastes … it may become necessary to change the conceptual framework of our theory” (von Weizsäcker, 1971, p. 346). He (2005, p. 2) formulates the challenge succinctly:

After all, preferences of individual agents are the basic measuring rod of economic welfare, of the performance generated in an economic system. How can we evaluate an economic system with a measuring rod that itself changes with the system?

Von Weizsäcker has developed a rich conceptual answer to this question over a series of articles (1971, 2002, 2005, 2013a, 2013b, 2014a, 2014b, 2015). A core idea of his approach is that an answer to the aforementioned question can be given if “some ‘laws of motion’ of preferences” can be identified according to which preferences change along non-circular “improvement paths” (2005, p. 4).Footnote 3 He calls preferences that follow such improvement paths adaptive preferences. This paper does not aim to cover von Weizsäcker’s work in its entirety but focus on the more recent article The Normative Co-Evolution of the Market Economy and Democracy (2014a).Footnote 4 This article is particularly relevant for the argument of this paper since it directly addresses some of the blind spots of Vanberg’s CPE program, in particular its neglect to consider the inherent status quo tendency of preference adaptation, social influences on preference formation processes, and structural conditions for individual autonomy.

3.1 Two crucial aspects of preference endogeneity: adaptation and interpersonal influences

The first refinement of the CPE framework is von Weizsäcker’s insight that a core feature of preference endogeneity is the logic of adaptivity. Preference adaptivity simply means that people adjust their preferences to given situational or social circumstances. According to von Weizsäcker (2013b), preference adaptation is a realistic theoretical assumption that helps explain widespread behavioral patterns observed in practice, both on an intrapersonal level (in the form of habit formation) and on an interpersonal level (in the form of social emulation).Footnote 5Adaptivity of preferences implies a strong adherence to the status quo. Economic reasons for preference adaptivity are switching costs (moving away from the status quo often entails substantial upfront costs) and information asymmetries (due to information uncertainties of alternatives many prefer the status quo). The existence of adaptive preferences means that policies will likely be endorsed by people’s ex post preferences if they are also endorsed by those people’s ex ante preferences.Footnote 6 In other words, people value policies more highly if they have endorsed them in the past. This pattern follows the idea of habit formation in private life, i.e., the tendency to exhibit behaviors that one has engaged in the past.

The second important aspect of preference endogeneity has to do with interpersonal influences on individual processes of preference formation. According to von Weizsäcker, people influence each other’s preferences constantly. This is not just true for preferences for market goods, but also for preferences for ideas, moral values, or aesthetic ideals (2014a, p. 17). In fact, von Weizsäcker states that “the interpersonal influence on individual preferences and values is the rule rather than the exception” (2014a, p. 18). In the context of interpersonal influences, von Weizsäcker differentiates between the influencer person B and the influenced person A. B can intentionally or unintentionally influence A. Importantly, B’s influence on A can be dominant or it can stand in competition with other influences (Fig. 1). These aspects are summarized in the following table with examples for each scenario:

Fig. 1
figure 1

Four classes of interpersonal influences (adjusted from von Weizsäcker, 2014a)

Full size image

3.2 The crucial role of competition

Given the ubiquity of interpersonal influences on individuals’ preference formation processes, von Weizsäcker (2014a), p. 18) asks: “What influences on the preferences of others are suitable for legitimizing the resulting preferences in society?” With his answer, von Weizsäcker aims to stay within the boundaries of a liberal society and normative individualism. He (2014a, p. 19) states that economists

cannot subject interpersonal influences on the preferences to an extensive ‘quality control’ by the state. On the other hand, it is also problematic to allow any kind of manipulation of citizens by other citizens.Footnote 7

According to von Weizsäcker, the principle of individual autonomy is central to the idea of a liberal society. It based on the assumption that adult citizens typically know best what actions contribute to their individual welfare. Yet, von Weizsäcker (2014a), p. 19) points out that in a world of ubiquitous interpersonal influences, competition among various influences is essential if economists want to legitimately use the resulting preferences as inputs for normative assessments of different social states:

If one wishes to maintain this principle of private autonomy even in the case of the universality of the interpersonal influences on preferences and values being accepted as fact, then one arrives at the competition of the influences on the preferences, or else at the competition of opinions as a criterion for the social legitimation of these influences.

The crucial point is that economists ought to avoid using preferences as normative inputs that are the result of influence monopolies:

Interpersonal influences on preferences are usually not an obstacle to acknowledging citizens’ preferences as a legitimate benchmark in the welfare of these citizens, as long as various influencers are in competition with each other while influencing the citizens, i.e., as long as the citizen is not subject to an influence monopoly. (Ibid.).

Von Weizsäcker concedes that individual autonomy is bounded in a world of interpersonal influences, but it is higher when there is competition among influences.Footnote 8 In taking this argumentative route, von Weizsäcker conceptualizes individual autonomy as a matter of degree, i.e., as a phenomenon that can be more or less present. Taking the issue of preference endogeneity seriously, economists should not simply assume consumer and citizen sovereignty but ask what the social conditions are (in the form of competition among influences) that diminish or enhance personal autonomy. Like in Vanberg’s CPE framework, von Weizsäcker considers information provision as a crucial condition for autonomous choice (2014a, p. 9). Yet, von Weizsäcker’s analysis of preference endogeneity goes deeper. Acknowledging the idea that preferences are “subject to what happens in society” (ibid.) opens up the possibility “to search for hidden constraints, in the sense of ‘Sozialkritik’. These constraints might be concealed behind the supposedly free decision of the citizen” (ibid.).Footnote 9

The importance of this statement cannot be underestimated. It means that economists should not take individual choices at face value in their normative analysis (as it is more or less done in CPE’s choice individualism), but must define circumstances under which ‘hidden constraints’ are reduced so that choices are free (or freer) and autonomous (or more autonomous). Von Weizsäcker highlights that this search for ‘hidden influences’ on people’s preference formation processes “should be in the spirit of ‘Piecemeal Engineering’ in concrete and local circumstances and abstain from propagating a general ‘cultural revolution.’” (2014a, p. 21). While emphasizing ‘Piecemeal Engineering’ on a pragmatic implementation level, the decisive theoretical point is that preferences of citizens are regarded as the legitimate normative guidelines for welfare economists’ analyses only if certain social-structural conditions are fulfilled. The next section will discuss the contours of those structural conditions.

3.3 The twin ideas of deliberative democracy and competitive market economy

According to von Weizsäcker, a liberal democratic order will be more psychologically stable (i.e., its citizens are more likely to support the implications of its socio-political structure), when its economic system is based on a competitive market economy. History of capitalism suggests that a competitive market economy enables material progress which in turn contributes to what von Weizsäcker calls “improvement sequences” (2014a, p. 12). Improvement sequences built on inclusive material growth follow the principle of non-circularity: after a series of changes from state x0 to xT, people do not want to go back to x0 at point T. Improvement sequences mean that citizens do not want to go back to ‘square one’ when thinking about the value of incremental institutional change.

Another reason why a competitive market economy leads to a higher stability of the liberal democratic order is the adaptivity of people’s preferences. In politics, with its centralized decision-making structure, a majority is needed to overcome the status quo. However, if we assume adaptive preferences of the majority, a deviation of the status quo is unlikely. This means that impulses for change in the form of preference entrepreneurship often come from non-political actors. A ‘total’ democracy, in which all important decisions would be made according to majority rule, would likely be a stagnant, non-prosperous society due to the adaptive nature of the majority’s preferences. A paradigmatic example of preference entrepreneurship is Apple’s Steve Jobs whose radical product innovations from smartphones to smartwatches have contributed to material progress and increased consumer welfare. Such changes would likely not be possible in a centralized decision structure of politics where a majority with adaptive preferences might vote against the implementation of radical technological innovations.

At the same time, von Weizsäcker emphasizes the communicative structure of deliberative democracy as a crucial condition for accepting preferences of citizens as a legitimate input for economists’ welfare analysis. In an economy with endogenous preferences and interpersonal influences, the legitimacy of the resulting economic order presupposes deliberative democracy to ensure not just competition among products and services but also among cultural mental models that play out in markets and in politics. Von Weizsäcker highlights the crucial role of institutionalized processes of public reasoning for individual autonomy:

“The communication structure of a deliberative democracy is a condition for the preferences of citizens being accepted as a legitimate measure of citizen welfare. In this sense, democracy not only presupposes the market economy; in turn, the legitimacy of the market economy presupposes a deliberative democracy in the context of endogenous preferences and, in particular, interpersonal influences on preferences.” (2014a, p. 20).

Public discourse is not only important to counterbalance dominant interpersonal influences in the realm of cultural mental models, but it can also be the catalysator of change in a world of adaptive preferences and thus prevent further entrenchment of the power of a majority.

So far, the paper has argued that Vanberg’s defense of choice individualism and his advocacy of information provision underestimates the various challenges preference endogeneity poses for normative analysis in economics. Von Weizsäcker rightly identifies preference adaptation and interpersonal influences as core challenges of preference endogeneity. When discussing the necessary structural conditions for autonomous choice, von Weizsäcker defends the ideas of competition and public reasoning to overcome the status quo tendency of adaptive preferences and reduce the power of dominant interpersonal influences. Overall, von Weizsäcker acknowledges that individual autonomy is a matter of degree and highlights the important role of external structural conditions to increase autonomy. In doing so, von Weizsäcker’s approach can be considered an advancement of Vanberg’s CPE. However, as the next section will argue, von Weizsäcker’s neglects a discussion of conditions that foster individuals’ internal agentic capabilities which allow them to reflect upon their evolving preferences. To put it in evolutionary terms, von Weizsäcker rightly highlights the crucial role of variation in the social provision of products and cultural mental models, but he neglects a more intricate discussion of the crucial role of cognitive selection mechanisms.

4 Missing elements: process freedom and agentic capabilities

By focusing on the value of competition, von Weizsäcker’s approach emphasises the opportunity dimension of freedom, i.e., the presence of options among which one can freely choose. When preferences are endogenous to situational framing and the wider socio-cultural environment, increasing the number of choice options via competition likely increases individual autonomy. The reason is that, “[the] richer the set of opportunities from which a person has chosen his way of life, the more that way of life is his.” (Sudgen, 1998, p. 311). For instance, in the context of the subjugation of women, a situation with a larger opportunity set (when women can choose between being a housewife or entering the regular labor market) likely leads to a higher sense of autonomy compared with a situation with a smaller opportunity set (when women have few options outside family life) (Sugden, 2018, pp. 99–100).

Neither von Weizsäcker, nor Vanberg provides us with an explicit definition of autonomy. Since both of them take normative individualism as their starting point, within their frameworks autonomy cannot be defined in objective terms but must be understood subjectively. Following Sugden’s subjective definition, autonomy can be understood as the feeling of an agent to identify “with her own actions, past, present, and future” (2018, p. 106). An autonomous agent has a high degree of subjective perception of herself “as the cause of her own actions” (Sugden, 2019, 27). She takes responsibility for the consequences of her consequences (since the action are hers) and considers herself as the author of her own life. The crucial question is: Are opportunities to choose from enough to enable this sense of autonomy?

4.1 The limits of competition-only

Let us consider two alternative opportunity sets X and Y. X is a small town in New England and Y is a metropolis in California. There is more competition among interpersonal influences in Y than in X which means that Y offers more opportunities. For simplicity reasons, we assume that Y is a superset of X (Y ⊃ X). According to von Weizsäcker’s logic, a person being exposed to opportunity set Y with more competition should feel a higher sense of autonomy. Yet this might not be the case if we acknowledge that at a certain point more competition and larger choice sets can actually lead to a lower sense of individual autonomy (Dold & Rizzo, 2021). For instance, the inclusion of additional elements in the large opportunity set Y might make it likelier that individuals rely on defaults and random environmental cues to make decisions. In this case, it is conceivable that for a certain class of choices (e.g., among potential partners or different jobs) the individual’s sense of autonomy might be compromised in the more competitive setting. In other words, she might feel that the choices she makes are not fully hers. Of course, as the example of the subjugation of women from above illustrates, more competition can lead to a higher sense of autonomy, whereas situations with less competition and smaller opportunity sets can lead to a lower sense of autonomy (see O1-O4-diagonal in Fig. 2 below).

Fig. 2
figure 2

The Interplay of Autonomy and Competition (adapted from Dold & Rizzo, 2021)

Full size image

However, competition need not always be good for a person’s sense of autonomy. Behavioral economics suggests that there are many real-world cases for O3 where individuals are exposed to more competition (i.e., a larger opportunity set) but perceive a lower sense of autonomy. We can think of situations where an individual is overwhelmed by the sheer amount of additional options (choice overload), when additional choice options that look particularly attractive in the moment obscure the long-term value of some other elements in the choice set (obfuscation), or when individuals rely on random contextual cues to reduce decision costs (context effects). In these cases, a person might make choices she cannot fully identify with and hence be interested in restricting the size of the opportunity set to increase her sense of autonomy. While a one-time decision in a large opportunity set scenario might not lead to a lower sense of autonomy, the effect might be more pronounced when a person reflects on a series of choices and realizes how much her preference formation process was shaped by the context despite the higher degree of competition (Dold & Rizzo, 2021, pp. 368–369). Individuals interested in both a sense of autonomy and freedom of choice (more opportunities) will find subjective ways to trade these values off against each other. While O1 dominates O4, it is not clear how individuals will rank O2 and O3 respectively. It is conceivable that some individuals will prefer O2 (low degree of competition, high level of autonomy) to O3 (high degree of competition, low level of autonomy) if they value the increase in autonomy due to less competition in O2 more than a potential increase in opportunities at a lower level of autonomy in O3. Summing up, people might not only have an interested in choosing among many opportunities, but also among different menus of opportunities since menus differently affect people’s sense of autonomy and shape their preference formation processes.

4.2 Process freedom

The analysis of the previous section suggests that people may not only care about opportunity freedom (the availability of achievable options) but also process freedom (a person’s capacity to control the choice process). Sen (2004, p. 10) points out that both of these aspects of freedom are valuable:

Freedom can be valued for the substantive opportunity it gives to the pursuit of our objectives and goals ... [so that] the focus is not directly on what the processes involved happen to be, but on what the real opportunities of achievement are for the persons involved. This ‘opportunity aspect’ of freedom can be contrasted with another perspective that focuses in particular on the freedom involved in the process itself (for example, whether the person was free to choose himself, whether others intruded or obstructed, and so on). This is the ‘process aspect’ of freedom.

The process and opportunity dimensions of freedom can complement each other. For instance, when a professor has the opportunity to work for a number of universities, she might feel in charge of the decision-making process since she can compare the relative merits of the choice options better in a larger choice set scenario. However, the process and opportunity dimensions of freedom can also diverge. For instance, a young assistant professor might have the power to achieve certain opportunities (e.g., publish a paper in journal edited by her co-authors and friends) but lack the means to critically reflect upon such actions due to publication and peer pressure. Having process freedom would mean that the young assistant professor’s preference formation process is not dominated by social and cognitive factors that diminish the individual’s sense of autonomy.

When considering the process dimension of freedom, Sen emphasises the importance of reasoned scrutiny that precedes choice: “What is needed is not merely freedom and power to act, but also freedom and power to question and reassess the prevailing norms and values” (Dreze and Sen, 2002, p. 258). Reasoned scrutiny of prevailing norms and values is a crucial feature of feeling in charge of one’s decisions. Conversely, preference adaptation in the form of unconsidered internalization of cultural mental models can be seen as a prime obstacle to autonomy. Reasoned scrutiny is not only important to counterbalance processes of preference adaptation and social manipulation, but also for ensuring that “[the] agent’s decision is not for no reason, based on a whim or impulse, but is for some reason or to achieve some goal” (Crocker & Robeyns, 2010, p. 81). As Sen points out:

The role of preference revision and reform [is an essential] part of the freedom of living. … The scrutiny and cultivation of preferences—and the freedom to be able to do that (whether or not one actually does it)—can be quite relevant to the assessment of a person’s overall opportunities. (Sen, 2004, p. 618).

Applied repeatedly and iteratively, reasoned scrutiny of one’s preference formation process that incorporates affective feedback leads to a more integrated identity and higher levels of subjective well-being. Reasoned scrutiny helps ensure that people are more satisfied with the preferences and habits they end up developing (Fabian & Dold, 2022). In this sense, process freedom is instrumentally valuable as a means to good well-being consequences. If people feel that they are in charge of their decision-making process, i.e., they feel competent when scrutinizing, forming, and deciding upon their preferences, their actions are more likely to result in well-being achievements and outcomes they can fully identify with (Crocker & Robeyns, 2010, p. 83).

Process freedom requires a person to have agentic capabilities, such as the ability to assess a multitude of choice options and to form preferences in a self-reflective way, thereby selecting among competing opportunities the option one can identify with and take responsibility for (Dold & Lewis, 2023). Hence, agentic capabilities reflect the “conditions under which people acquire the sense of interest on which they act.” (Hargreaves Heap, 2013, p. 995). Emphasising the importance of internal agentic capabilities besides external factors such as information provision and competition is consistent with the liberal view that “whatever action people take, they should feel they own it in the sense that they have had the resources to reflect on what preferences to hold and how to act on them” (ibid.).

The idea of the central role of agentic capabilities for autonomous choice is supported by Buchanan’s article Natural and Artifactual Man (1979 [1999]) in which he explicitly endorsees the idea of endogenous preferences. In the article, Buchanan identifies imagination and valuation as core agentic capabilities when individuals are faced with expanding opportunity sets (Dold, 2018). Buchanan does not only emphasize the importance of competition and the availability of a wide range of opportunities, but concedes that “the ranking of prospects requires valuation” and that society should “provide persons with both an array of imagined prospects and some means of valuation” (Buchanan, 1979 [1999], p. 254). Summing up, for individuals to develop a robust sense of autonomy they may need agentic capabilities. These enable a person to feel in charge of her choices and to be able to make competent choices that cohere with her overall conception of who she is or wants to become.

5 Policy considerations

This paper has argued that neither information provision, as in Vanberg’s framework of choice individualism, nor the implementation of competition in the market and political sphere, as in von Weizsäcker’s framework, are likely sufficient to ensure individual autonomy when preferences are endogenous. Taking the challenge of preference endogeneity for discussions of individual autonomy seriously, the paper has argued that economists may need to consider process aspects along opportunity aspects of freedom. The paper has argued that a joint concern for process and opportunity freedom provides a sensible condition for when preferences are more likely to be indicative of individual welfare and therefore constitutes a better normative benchmark for economists’ evaluation of social states. This final section aims to provide a brief discussion of some implications of this perspective for normative discourse and policymaking.

First, behaviourally-informed normative economics that takes people’s autonomy concerns seriously “would seem naturally to be concerned with the conditions (e.g., the educational system, the media, the family, vibrancy of the arts world) that support reflection on what preferences to hold” (Hargreaves Heap, 2013, p. 995). In doing so, this perspective emphasizes social-structural conditions for autonomous individual choice. This follows Sen (2004, p. 625) who notes that “personal process concern … must include consideration of the way in which the person’s life is affected by general processes operating in society.” Crucially, as von Weizsäcker has rightly noted, behaviourally-informed normative economics can address the conditions under which people make decisions and form preferences (e.g., push back against forms of social domination and manipulation) without being committed to any view about what those preferences ought to be.

Second, an acknowledgment of preference endogeneity means that policies may need to be less focused on concrete outcomes and rather focus on input factors, such as agentic capabilities, which people may or may not possess when reflecting and acting upon their evolving preferences (Fig. 3). Hence, highlighting the importance of process freedom may shift the policy focus from choice outcomes to input factors that contribute to cognitive process facilitation (see figure below).

Fig. 3
figure 3

(adapted from Sher et al., 2022)

Two aims and three intervention targets in choice architecture.

Full size image

In libertarian paternalism, the currently dominant approach in behavioural public policy, policymakers aim to facilitate choice outcomes by means of nudges, i.e., subtle changes in the choice architecture such as changes in default rules or rearrangements of items on a menu. In doing so, nudges typically do not enhance people’s reflective capacities but exploit their cognitive biases to steer them toward a predefined choice goals (such as ‘more savings’ or ‘healthier food’). In contrast, the approach this paper advocates for would seek to implement policies that facilitate people’s choice processes without prejudging the choice outcomes. Choice process interventions would aim both at the enhancement of the quality of informational inputs (e.g., by reducing ‘sludge’) and people’s cognitive operations (e.g., by means of ‘boosts’ that target people’s agentic capabilities). Sludge is information processing friction caused by, among other things, overwhelming, poorly formatted, nontransparent, misleading, or hard-to-find information. Sludge increases decision costs and makes life difficult to navigate. Examples of sludge can be complex interfaces whose use of terms knowingly confuses their users; price information that shrouds the actual costs of products; or choice options that are hard to opt out (such as online subscriptions). If there is sludge, many consumers do not even begin to engage with the information provided. Sludge audits can help reduce such information processing frictions and thereby enhance consumer autonomy (Sunstein, 2020). Boosts, on the other hand, are “interventions that make it easier for people to exercise their own agency by fostering existing competences or instilling new ones.” (Hertwig & Grüne-Yanoff, 2017, 974). To understand boosts, it is helpful to contrast them with nudges. A paradigmatic nudge in the realm of financial decision-making is the idea of Save More Tomorrow (SMT). The SMT nudge exploits people’s present bias and inertia by encouraging them to sign up to a financial plan that links their increases in retirement contributions to future salary increases. In contrast, a boost intervention would aim to de-bias the decision-maker by directly targeting their agentic capabilities, e.g., by providing training in basic financial heuristics (such as a simple 1/N diversification strategy).

Third, and finally, there might also be limits to ‘planning’ for agentic capabilities. John Stuart Mill’s work is particularly instructive in this regard. According to Mill (1859 [1998]) a person obtains individuality—they develop their own preferences—when they have been exposed to ‘experiments in living.’ Such experiments allow people to develop agentic capabilities by being exposed to other viewpoints and novel situations. Yet, Mill also highlights the role of the individual in the acquisition of agentic capabilities. According to Mill, the process of individuation requires continuous active choices on behalf of the individual. Mill (1859 [1998], p. 123) states:

The human faculties of perception, judgment, discriminative feeling, mental activity, and even moral preference, are exercised only in making a choice. He who does anything because it is the custom, makes no choice. He gains no practice either in discerning or in desiring what is best. The mental and moral, like the muscular powers, are improved only by being used.

The Millian perspective suggests that it may be misleading to focus solely on top-down policies and the idea, prevalent in much of the nudging literature, of helping people realize the subset of their preferences that remains relatively stable over time. Instead, Mill regards the self-directed development of new preferences as an indispensable part of autonomy and as an essential source of well-being. He emphasizes that no third party, such as experts or the social group one belongs to, can fully replace these internal processes of preference development:

“He who lets the world, or his own portion of it, choose his plan of life for him, has no need of any other faculty than the ape-like one of imitation.” (1859 [1998], p. 124).