A Defence of Informed Preference Satisfaction Theories of Welfare

Abstract

This article defends informed preference satisfaction theories of welfare against the most influential objections put forward in the economic and philosophy of science literatures. The article explicates and addresses in turn: the objection from inner rational agents; the objection from unfeasible preference reconstruction; the objection from dubious normative commitments; the objection from conceptual ambiguity; and the objection from conceptual replacement. My defence does not exclude that preference satisfaction theories of welfare face significant conceptual and practical challenges. Still, if correct, it demonstrates that philosophers/welfare economists are justified in relying on specific versions of such theories, namely informed preference satisfaction theories.

1. Introduction

Theories of welfare are commonly classified into mental state theories of welfare, which hold that individuals are well-off to the extent that they experience specific kinds of mental states (e.g., Clark et al. 2018, on happiness; Feldman 1997, on pleasure), preference satisfaction theories of welfare (henceforth, PSTW), which hold that individuals are well-off to the extent that their own preferences are satisfied (e.g., Bernheim 2009; Ferreira 2023), and objective list theories of welfare, which hold that individuals are well-off to the extent that they have certain goods/experiences (e.g., health, education, friendship) irrespective of whether they experience specific kinds of mental states or satisfy their preferences (e.g., Fletcher 2013; Nussbaum 2000). ¹ Philosophers/welfare economists frequently assume that individuals’ preferences can be reliably inferred from their choices and often rely on PSTW (e.g., Adler and Fleurbaey 2016; Angner 2016; Sobel 1998; Sumner 1996, ch. 5). In recent years, however, several prominent authors have argued that PSTW fail to provide a plausible theory of welfare. In this article, I develop and support a qualified defence of PSTW against the most influential objections put forward in the economic and philosophy of science literatures. My main claim is that although PSTW face significant conceptual and practical challenges, specific versions of PSTW – namely, informed PSTW – are more plausible than their critics maintain, and philosophers/welfare economists are justified in relying on such versions of PSTW. ²

The article is structured as follows. Section 2 outlines the main tenets of PSTW and distinguishes between the three main proffered versions of PSTW, namely actual PSTW, informed PSTW, and ideal PSTW. Sections 3–7 defend informed PSTW against five prominent objections. More specifically, I explicate and address in turn: (i) the objection from inner rational agents (e.g., Infante et al. 2016a; Sugden 2018, ch. 4–5); (ii) the objection from unfeasible preference reconstruction (e.g., Infante et al. 2016b; Rizzo and Whitman 2020, ch. 6–7); (iii) the objection from dubious normative commitments (e.g., Hausman and McPherson 2009; Kraut 2007, part II); (iv) the objection from conceptual ambiguity (e.g., Hausman 2024; Lecouteux 2022); and (v) the objection from conceptual replacement (e.g., Levy and Glimcher 2012; Thoma 2021a).

Before proceeding, the following three preliminary remarks are worth making. First, the objections I address target the main tenets of informed PSTW, which concern the existence of informed preferences, the possibility of reconstructing these preferences, and the normative/evaluative significance of such preferences. These objections do not exhaust the set of critical issues faced by informed PSTW (e.g., Sugden 2008; Whitman and Rizzo 2015, holding that adopting informed PSTW leads philosophers/welfare economists to endorse unjustified paternalistic interventions; Fumagalli 2024, for a reply). Still, as I illustrate below, those objections target the main respects in which the plausibility of informed PSTW has been called into question in the recent economic and philosophy of science literatures. I take such objections to be especially interesting to philosophers/welfare economists since they encompass the major bones of contention between the proponents and the critics of informed PSTW and highlight the most pressing challenges faced by PSTW more generally.

Second, PSTW are not the only approach that aims to ground reliable and informative welfare evaluations on information concerning individuals’ preferences. Still, PSTW differ from several other preference-based approaches to welfare evaluations in that PSTW take preference satisfaction to constitute welfare rather than merely provide evidence for welfare (e.g., Hausman and McPherson 2009, 1–2, on the evidential account of welfare, which holds that ‘if individuals seek to benefit themselves and are good judges of what is good for them, then […] their preferences will be reliable indicators of what is good for them’). I do not aim in this article to assess the comparative strengths and limitations of different preference-based approaches to welfare evaluations. However, I shall expand on preference-based approaches other than informed PSTW when consideration of such approaches directly bears on my defence of informed PSTW (e.g., Section 5 on the evidential account of welfare). ³

And third, my defence of informed PSTW primarily focuses on individual welfare evaluations rather than social welfare evaluations since social welfare evaluations raise additional complications that are orthogonal to the merits of informed PSTW (e.g., Adler 2012, ch. 3; Fleurbaey 2012, on various epistemic challenges faced by attempts to ground interpersonal comparisons of welfare; Adler 2019, ch. 3–4; Fleurbaey and Maniquet 2011, ch. 2–4, on various normative challenges faced by attempts to determine what weights should be ascribed to different individuals’ welfare in interpersonal aggregations of welfare). Still, I shall comment in various places on the applicability of informed PSTW to social welfare evaluations (e.g., Sections 3–7 on the evaluation of policies’ welfare implications). ⁴

2. Preference satisfaction theories of welfare

According to PSTW, the satisfaction of individuals’ preferences constitutes individuals’ welfare, i.e. makes individuals better off than they would be in otherwise identical situations where their preferences are not satisfied (e.g., Fumagalli 2021; Hausman 2012, ch. 7–8). In particular, an individual’s preferences for some state of affairs count as satisfied if such state of affairs occurs (e.g., Hausman 2010, 326; also Griffin 1986, ch. 1; Sumner 1996, ch. 1). Individuals may derive feelings of pleasure or satisfaction from knowing that their preferences are satisfied. Still, on PSTW, preference satisfaction does not have to involve any feelings of pleasure or satisfaction (e.g., Hausman and McPherson 2009, 13; Kraut 2007, 98–99) and may constitute welfare irrespective of whether it involves such feelings (e.g., Hausman and McPherson 2009, 10, holding that ‘there is only [a] contingent connection between the satisfaction of a preference and the satisfaction [felt by] a person’; also Arneson 1999, 123; Rabinowicz and Österberg 1996, 2). ⁵

Three main versions of PSTW have been articulated in the specialized literature. Actual PSTW take individuals’ welfare to be constituted by the satisfaction of their actual preferences, i.e. the preferences individuals happen to have in the examined choice settings (e.g., Gul and Pesendorfer 2008, 24). For their part, informed PSTW take individuals’ welfare to be constituted by the satisfaction of their informed preferences, i.e. the preferences individuals are able to form on the basis of accurate information and considerate judgements concerning their choice options/circumstances (e.g., Griffin 1986, ch. 1–2). Still differently, ideal PSTW take individuals’ welfare to be constituted by the satisfaction of their ideal preferences, i.e. the preferences individuals would counterfactually have ‘if they had complete information [concerning their choice options/circumstances and] unlimited cognitive abilities’ (Sunstein and Thaler 2003, 1162; also Harsanyi 1982, 55, on the preferences individuals would have if they ‘had all the relevant factual information [and] were in a state of mind most conducive to rational choice’). ⁶

The distinction between actual PSTW, informed PSTW, and ideal PSTW categorizes the proffered versions of PSTW into three exclusive sets. These versions of PSTW require preferences to meet dissimilar conditions to qualify as actual preferences, informed preferences, and ideal preferences, respectively. I shall expand on these dissimilarities in Sections 3–7. For now, I note that – contrary to actual PSTW – both informed PSTW and ideal PSTW impose at least two conditions on individuals’ preferences, namely information (epistemic) conditions, which concern the extent to which individuals’ preferences are grounded on accurate information concerning individuals’ choice options/circumstances, and consistency (structural rationality) conditions, which concern the extent to which individuals’ preferences fit specific consistency requirements (e.g., transitivity). The idea is that only some of individuals’ preferences are such that their satisfaction constitutes individuals’ welfare and that only preferences which satisfy specific information conditions (e.g., accurate information about the available choice options) and specific consistency conditions (e.g., transitivity) are plausibly taken to belong to such a set of welfare-relevant preferences (e.g., Griffin 1986, ch. 1–2; Fumagalli 2025). ⁷

More specifically, ideal PSTW impose rather demanding information and consistency conditions on preferences, in that they hold that only fully informed and consistent preferences are such that their satisfaction constitutes welfare (e.g., Harsanyi 1982, 55; Sunstein and Thaler 2003, 1162). For their part, informed PSTW impose less demanding information and consistency conditions on preferences, in that they allow that the satisfaction of incompletely informed and partly inconsistent preferences may constitute welfare (e.g., Bernheim 2021, 390–92; Fumagalli 2024, 89–91). Distinct versions of informed PSTW impose different information and consistency conditions on preferences (e.g., Griffin 1986, ch. 1). These differences by no means justify taking any information and consistency conditions to reliably track individuals’ welfare (e.g., Sections 3–4 on the inadequacy of various information and consistency conditions). Still, those differences are not inherently problematic for the proponents of informed PSTW. For both information and consistency are plausibly taken to admit of degrees, and the specification of the information and consistency conditions presupposed by informed PSTW may justifiably vary across choice settings (e.g., how much information individuals must possess for their preferences to qualify as informed may justifiably vary depending on what individuals are involved and what choices they face). I shall explicate such differences in Sections 3–7. ⁸

According to the critics of PSTW, neither actual PSTW nor informed/ideal PSTW withstand scrutiny. The critics’ case against actual PSTW can be explicated as follows. Let us call those actual preferences whose satisfaction is plausibly taken to constitute welfare (if any) actual preferences*. Philosophers/welfare economists can identify actual preferences* only if individuals’ actual preferences meet stringent information conditions (e.g., accurate information about the available choice options) and consistency conditions (e.g., transitivity). However, individuals’ actual preferences frequently fail to meet these conditions (e.g., Sugden 1991, on violations of transitivity; Hausman 2011, on cases where individuals’ actual preferences rest on inaccurate information about the available choice options). Moreover, individuals’ actual preferences often track factors that appear to be prudentially irrelevant (e.g., Camerer and Loewenstein 2004, on cases where individuals’ actual preferences depend on frames) or even hamper what most theories of welfare regard as individuals’ welfare (e.g., Hausman and McPherson 2009, on cases where individuals prefer options that hamper their welfare because they mistakenly believe that such options enhance their welfare; Stoljar 2014, on cases where individuals prefer options that hamper their welfare as a result of adaptation). For these reasons, the critics of actual PSTW go, the satisfaction of individuals’ actual preferences cannot be plausibly taken to constitute individuals’ welfare. ⁹

The critics’ case against informed/ideal PSTW can be explicated as follows. ¹⁰ Let us call those informed/ideal preferences whose satisfaction is plausibly taken to constitute welfare (if any) informed/ideal preferences*. Informed/ideal PSTW can accommodate the fact that individuals’ actual preferences are often inconsistent, ill-informed, and track factors that appear to be prudentially irrelevant/detrimental. For as leading critics of PSTW acknowledge, one may plausibly ascribe to individuals informed/ideal preferences* in several cases where individuals’ actual preferences are inconsistent, ill-informed, and track factors that appear to be prudentially irrelevant/detrimental (e.g., Hausman and McPherson 2009, 11, holding that informed/ideal PSTW ‘resolve […] most of the difficulties facing the actual preference-satisfaction view’ since a person’s actual preferences often fail to be ‘informed’ ‘and so satisfying [such preferences] would not make the person better off’; also Loeb 1995, 1; Rosati 2006, 63; Sumner 1996, ch. 5). However, it is difficult to determine what notion of informed/ideal preferences should ground welfare evaluations unless one makes substantive normative/evaluative assumptions (e.g., McQuillin and Sugden 2012, 560, claiming that the concepts of ‘complete information’ and ‘unlimited cognition’ figuring in ideal PSTW are ‘inescapably normative’). Moreover, different approaches have been developed to reconstruct individuals’ informed/ideal preferences*, and different approaches classify different subsets of preferences as informed/ideal preferences* (e.g., Dold 2018; Whitman and Rizzo 2015, for illustrations). In fact, reconstructing some individuals’ informed/ideal preferences does not per se enable philosophers/welfare economists to reliably assess these individuals’ welfare (e.g., Sugden 2018, ch. 4, on putative cases where individuals’ choices reveal context-dependent informed/ideal preferences). For these reasons, the critics of informed/ideal PSTW go, the satisfaction of individuals’ informed/ideal preferences cannot be plausibly taken to constitute individuals’ welfare. ¹¹

In the following sections I argue that the proffered criticisms of PSTW cast doubt on both actual PSTW and ideal PSTW, but fail to undermine informed PSTW. In particular, I shall defend informed PSTW against five prominent objections, namely: the objection from inner rational agents (Section 3); the objection from unfeasible preference reconstruction (Section 4); the objection from dubious normative commitments (Section 5); the objection from conceptual ambiguity (Section 6); and the objection from conceptual replacement (Section 7). ¹²

3. Objection from inner rational agents

The objection from inner rational agents holds that informed PSTW do not withstand scrutiny because individuals cannot be plausibly taken to have well-informed and consistent informed preferences*. The objection proceeds as follows. Individuals’ informed preferences can be regarded as either actual attitudes or merely hypothetical attitudes. Now, if individuals’ informed preferences are regarded as actual attitudes, then the claim that individuals have ‘inner rational agents’ with well-informed and consistent informed preferences* ‘lacks psychological foundations’ (Sugden 2018, 13, italics added). For ‘there is no general reason’ to think that ‘inner rational agents’ with well-informed and consistent informed preferences* ‘exist at all’ (Infante et al. 2016b, 34; also Infante et al. 2016a, 22). Conversely, if individuals’ informed preferences are regarded as merely hypothetical attitudes, then these preferences lack sufficient connection to individuals’ welfare to ground reliable and informative welfare evaluations. For what one would prefer under hypothetical circumstances may be rather uninformative about her welfare (e.g., Cowen 1993, 265, holding that ‘a self with radically different brain endowments and capacities […] cannot judge my welfare [because such self] is a different individual altogether’).

This objection correctly notes that some versions of informed PSTW presuppose (rather than show) that individuals have a set of well-informed and consistent informed preferences* (e.g., Infante et al. 2016a, for illustrations). Still, there are at least two reasons to doubt that the objection undermines informed PSTW. First, informed PSTW do not rest on the assumption that individuals have inner rational agents with informed preferences*. In particular, the proponents of informed PSTW may provide detailed specifications of the conditions under which individuals are plausibly ascribed informed preferences* without having to posit any inner rational agents having such preferences* (e.g., Hausman 2016). For informed PSTW’s information and consistency conditions do not concern whether individuals’ preferences are formed via any particular psychological process. And the proponents of informed PSTW can determine what preferences meet such conditions without having to posit any inner rational agents (e.g., Beck 2023; Fumagalli 2024). ¹³

And second, the proponents of informed PSTW can identify preferences that both meet the information and consistency conditions presupposed by informed PSTW and have sufficient connection to individuals’ welfare to ground reliable and informative welfare evaluations (e.g., Bernheim 2021; Fumagalli 2025, on various sets of well-informed and transitive preferences). To be sure, the fact that an individual’s preferences meet some information and consistency conditions does not per se imply that satisfying such preferences is plausibly taken to constitute the individual’s welfare. For not all information and consistency conditions are plausibly taken to reliably track individuals’ welfare (Section 2; also Fumagalli 2021). Still, the information and consistency conditions presupposed by leading versions of informed PSTW (e.g., transitivity, accurate information about the available choice options) provide a reliable criterion for reconstructing informed preferences*. For satisfying well-informed/consistent preferences tends to yield individuals higher welfare-relevant payoffs than satisfying ill-informed/inconsistent preferences (e.g., Beshears et al. 2008, on cases where satisfying ill-informed preferences prevents individuals from achieving their own welfare-related goals; Gustafsson 2022, sec. 4, on cases where satisfying intransitive preferences makes individuals vulnerable to sure loss). ¹⁴

A critic of informed PSTW may object that informed PSTW evaluate individuals’ welfare ‘relative to the preferences that [individuals] would have revealed if not subject to reasoning imperfections’, and so implicitly presuppose that individuals have well-informed and consistent latent preferences*, i.e. preferences ‘that are formed within the minds of individual[s and] do not correspond directly with objective properties of the external world’ (Infante et al. 2016a, 7 and 9, italics added; also Infante et al. 2016b, 33). However, the proponents of informed PSTW can ascribe individuals well-informed and consistent informed preferences* without presupposing that individuals have well-informed and consistent latent preferences*. To illustrate this, consider Bernheim and Rangel’s preference-based approach, which aims to reconstruct a range of informed preferences* in settings where individuals’ choices depend on ancillary conditions, i.e. ‘feature[s] of the choice environment that may affect behaviour, but [are] not taken as relevant to [welfare]’ (2009, 55). ¹⁵

Bernheim and Rangel’s approach relies on the notion of ‘unambiguous choice’ as its welfare criterion. The idea is that ‘one alternative is unambiguously superior to another if and only if the second is never chosen when the first is available’ to individuals (Bernheim 2016, 15). Conversely, when individuals’ choices between two options vary across ancillary conditions, one should regard it as indeterminate which option enhances individuals’ welfare unless the observed choices result from demonstrable mistakes, i.e. are ‘predicated on a characterization of the available options […] that is inconsistent with the information available’ to individuals and ‘there is some other option in the opportunity set that [individuals] would select [in the absence of such] characterization failure’ (2016, 48). According to some critics, Bernheim and Rangel’s approach presupposes that individuals have ‘a neoclassical agent deep inside that is struggling to surface’ (Rizzo and Whitman 2020, 80; also Sugden 2018, 57). However, the approach does not assume ‘a context-independent objective function […] defined over a domain encompassing all the options of potential interest’ (Bernheim 2021, 392). In particular, the approach does not define mistakes in terms of divergences between choices and latent preferences*, and ‘does not assume that error-free choices reveal’ well-informed and consistent latent preferences* (392). In fact, Bernheim explicitly claims that individuals frequently ‘aggregate the many diverse aspects of [their] experience only when called to [choose]’ (2016, 20). ¹⁶

4. Objection from unfeasible preference reconstruction

The objection from unfeasible preference reconstruction holds that informed PSTW do not withstand scrutiny because philosophers/welfare economists cannot reliably reconstruct well-informed and consistent informed preferences*. The objection proceeds as follows. Suppose, for the sake of argument, that individuals can be plausibly taken to have well-informed and consistent informed preferences*. Even so, the assumption that philosophers/welfare economists ‘can reconstruct [these preferences*] is a mirage’ (Sugden 2018, 14, italics added). For the information and consistency conditions presupposed by informed PSTW frequently allow for different (and sometimes contradictory) reconstructions of individuals’ informed preferences* (e.g., Matson 2022). And apparent conflicts between preferences can typically be resolved in multiple ways (e.g., Whitman and Rizzo 2015, on the difficulty of identifying welfare-optimal rates of saving and intertemporal discounting). Hence, philosophers/welfare economists often ‘have no means of determining which of the conflicting preferences reflect [informed preferences*]’ (Rizzo and Whitman 2020, 75; also Dold 2018, 161).

This objection correctly notes that philosophers’/welfare economists’ attempts to reconstruct informed preferences* face significant epistemic and normative challenges (e.g., Pettigrew 2023, on the epistemic and normative assumptions required to establish whether correcting specific inconsistencies enhances individuals’ welfare). Still, there are at least two reasons to doubt that the objection undermines informed PSTW. First, philosophers/welfare economists can reconstruct informed preferences* in several cases where the involved individuals fail to exhibit well-informed and consistent actual preferences (e.g., Bernheim and Rangel 2009, for reconstructions of informed preferences* in settings where choices are affected by ancillary conditions; Salant and Rubinstein 2008, for reconstructions of informed preferences* in settings where choices are affected by frames). To be sure, philosophers/welfare economists may be unable to reconstruct informed preferences* in presence of widespread choice inconsistencies (e.g., Sugden 2018, 58; also Bernheim 2016, 60, conceding that his approach ‘may not be very discerning […] in settings where choice inconsistencies are pervasive’). Yet, individuals’ choice inconsistencies are rarely so widespread that they prevent philosophers/welfare economists from reconstructing informed preferences*. To illustrate this, consider situations where individuals make some intransitive choices. These choices complicate philosophers’/welfare economists’ attempts to reconstruct individuals’ informed preferences*, but do not generally prevent philosophers/welfare economists from reconstructing such preferences*. For philosophers/welfare economists are frequently able to reconstruct informed preferences* in presence of some intransitive choices based on the core of transitive choices made by the involved individuals (e.g., Nishimura 2018, 589–99, for reconstructions of informed risk/time preferences* based on the core of transitive choices made by individuals). And philosophers/welfare economists can often point to experimental evidence demonstrating that individuals tend to regard transitivity as normatively compelling and are willing to revise intransitive choices when they realize these choices’ intransitivity (e.g., Hands 2014, 401–2; Nielsen and Rehbeck 2022, 2237–39).

And second, philosophers/welfare economists can frequently rely on multiple sources of evidence to reconstruct informed preferences*. In fact, philosophers/welfare economists have grounded several reconstructions of informed preferences* on both choice-based sources of evidence (e.g., Bernheim and Taubinsky 2018, on information concerning individuals’ hypothetical choices; Ferreira 2023, on information concerning the choices individuals would repeat at the time of welfare evaluation) and non-choice-based sources of evidence (e.g., Arieli et al. 2011, on eye-tracking data showing whether individuals attend to the available choice options; Bernheim 2016, on factual questions with objectively verifiable answers showing whether individuals understand the examined choice problems). To be sure, philosophers’/welfare economists’ attempts to reconstruct informed preferences* typically depend on normative/evaluative presuppositions about the notion of welfare (e.g., Haybron and Tiberius 2015, 714–17). However, these dependences do not reflect limitations inherent in informed PSTW, but rather reflect the thickness of the notion of welfare, i.e. the fact that this notion involves both positive and normative/evaluative dimensions (e.g., Dold and Schubert 2018, 223–24) and that, as a result, welfare ascriptions typically rely on both positive and normative/evaluative presuppositions (e.g., Fletcher 2019, 703–4).

A critic of informed PSTW may object that distinct sources of evidence may ground conflicting reconstructions of informed preferences* and that the information and consistency conditions presupposed by informed PSTW do not enable philosophers/welfare economists to discriminate between such reconstructions, i.e. to determine which of the proffered reconstructions of informed preferences reliably track informed preferences* (e.g., Whitman and Rizzo 2015, 420–4). The idea is that philosophers/welfare economists frequently face substantial normative ambiguity and that the information and consistency conditions presupposed by informed PSTW do not enable philosophers/welfare economists to resolve such ambiguity (e.g., Berg and Gigerenzer 2010, 148–50, on putative cases where satisfying ill-informed and inconsistent preferences enhances individuals’ welfare).

However, the objection significantly overestimates the extent of normative ambiguity inherent in individuals’ preferences. For as noted in Section 3, satisfying well-informed/consistent preferences tends to yield individuals higher welfare-relevant payoffs than satisfying ill-informed/inconsistent preferences. Moreover, the information and consistency conditions presupposed by informed PSTW provide a reliable (though fallible) basis to resolve the normative ambiguity inherent in individuals’ preferences by discriminating between conflicting reconstructions of informed preferences*, or at least by narrowing down the set of plausible reconstructions of such preferences*. To illustrate this, consider situations where individuals exhibit varying willingness to pay for specific goods/experiences across multiple frames. This variability complicates philosophers’/welfare economists’ attempts to reconstruct informed preferences*, but does not per se prevent philosophers/welfare economists from reconstructing a range of informed preferences* by demarcating precise and plausible bounds for minimal and maximal willingness to pay for the examined goods/experiences (e.g., Bernheim 2016, 60–64; also Abrahamson 2024, 24–26, for additional illustrations of philosophers’/welfare economists’ ability to reconstruct informed preferences* in cases where the involved individuals exhibit context-dependent preferences). ¹⁷

5. Objection from dubious normative commitments

The objection from dubious normative commitments holds that informed PSTW do not withstand scrutiny because informed PSTW’s normative assumption that the satisfaction of informed preferences constitutes welfare is implausible. The objection proceeds as follows. Suppose, for the sake of argument, that individuals can be plausibly taken to have well-informed and consistent informed preferences. Assume further that philosophers/welfare economists can reliably reconstruct these preferences. Even so, the satisfaction of such preferences is not plausibly taken to constitute individuals’ welfare. For a given state of affairs is not good for one ‘simply because [one prefers] with proper information, and reflectively [such state of affairs] to occur’ (Kraut 2007, 118, italics added; also Scanlon 1998, 115). In particular, ‘it is one thing to determine what people’s [informed] preferences would be […] and it is a different thing to determine what is good for people’ (Hausman 2016, 30, italics added; also Hausman and McPherson 2009, 12, holding that ‘the fact that [one] prefers x to y does not make it the case that x is better for [her] than y, no matter what conditions one imposes on [her] preferences’).

This objection correctly notes that substantiating informed PSTW requires the proponents of informed PSTW to support informed PSTW’s normative assumption that the satisfaction of informed preferences constitutes welfare. Still, there are at least two reasons to doubt that the objection undermines informed PSTW. First, supporting informed PSTW’s normative assumption that the satisfaction of informed preferences constitutes welfare is less demanding than the objection seems to presuppose. To illustrate this, let us distinguish between fundamental constituents of welfare – i.e. non-instrumentally valuable goods/experiences whose constitutive relation with welfare grounds the constitutive relation (if any) between all other non-instrumentally valuable goods/experiences and welfare – and intermediate constituents of welfare – i.e. non-instrumentally valuable goods/experiences whose constitutive relation with welfare is grounded in the constitutive relation between fundamental constituents and welfare. Supporting informed PSTW’s normative assumption that the satisfaction of informed preferences constitutes welfare does not require the proponents of informed PSTW to provide an exhaustive specification of all (fundamental and intermediate) constituents of welfare (e.g., Rabinowicz and Österberg 1996, 8–12). In particular, one may consistently endorse informed PSTW and acknowledge the existence of multiple intermediate constituents of welfare. For the issue of whether a given good/experience is a constituent of welfare is conceptually distinct from the issue of whether the constitutive relation between this good/experience and welfare (if any) is grounded in the constitutive relation between some other goods/experiences and welfare. In fact, several versions of informed PSTW allow that the satisfaction of informed preferences may constitute welfare through dissimilar intermediate constituents of welfare in different contexts (e.g., footnote 5 on the possibility that the satisfaction of informed preferences may constitute welfare through the contingent link between the satisfaction of an individual’s informed preferences and the sense of satisfaction that the individual may derive from knowing that such preferences are satisfied).

And second, the satisfaction of preferences that meet the information and consistency conditions presupposed by leading versions of informed PSTW (e.g., transitivity, accurate information about the available choice options) can be plausibly taken to constitute individuals’ welfare (Sections 3–4). To be sure, one may point to several cases where philosophers/welfare economists disagree as to whether satisfying specific sets of informed preferences constitutes individuals’ welfare (e.g., Griffin 1986, ch. 1; Sumner 1996, ch. 5, on cases where individuals are unaware that their informed preferences are satisfied; Parfit 1984, 494–95; Scanlon 1996, 111, on cases where individuals’ informed preferences target states of affairs that seem unrelated to individuals’ own welfare). However, the existence of contested cases does not per se license scepticism about informed PSTW. For many cases are not contested (e.g., individuals are often aware of whether their informed preferences are satisfied; individuals’ informed preferences frequently target states of affairs related to what most theories of welfare regard as individuals’ own welfare). And most contested cases are contested because of the normative/evaluative complexity inherent in such cases rather than because of alleged shortcomings inherent in informed PSTW (e.g., Fumagalli 2022, 532–33; Sunstein 2015, 518–19). That is to say, adopting theories of welfare other than informed PSTW does not per se enable philosophers/welfare economists to avoid contested cases (e.g., Griffin 1986, 17; Keller 2009, 656). And the proponents of informed PSTW may consistently endorse dissimilar positions concerning the proffered contested cases (e.g., Hawkins 2019, on recent debate about cases where individuals are unaware that their informed preferences are satisfied; Heathwood 2019, on recent debate about cases where individuals’ informed preferences target states of affairs that seem unrelated to individuals’ own welfare). ¹⁸

A critic of informed PSTW may object that informed PSTW rest on unnecessary normative commitments since philosophers/welfare economists can ground reliable and informative welfare evaluations on information concerning individuals’ informed preferences without endorsing any theory of welfare (e.g., Hausman 2010, 341; Hausman and McPherson 2009, 16). The idea is that philosophers/welfare economists should retain informed PSTW’s aim to ground reliable and informative welfare evaluations on information concerning individuals’ informed preferences, but should relinquish informed PSTW’s assumption that the satisfaction of informed preferences constitutes welfare because there is an evidential (rather than constitutive) connection between the satisfaction of informed preferences and welfare (e.g., Scanlon 1998, 116–18). However, it is dubious that appealing to this purported evidential connection undermines the justifiability of relying on informed PSTW. To illustrate this, consider the so-called evidential account of welfare, according to which ‘if individuals seek to benefit themselves and are good judges of what is good for them, then […] their preferences will be reliable indicators of what is good for them […] regardless of what theory of welfare one accepts’ (Hausman and McPherson 2009, 1–2, italics added; also Hausman 2012, 89).

The evidential account has gained significant prominence among the proponents of preference-based approaches in the recent economic and philosophy of science literatures (e.g., Beck 2023). Still, it is hard to establish whether the satisfaction of preferences that meet the conditions posited by the evidential account provides reliable evidence for welfare unless one relies on specific theories of welfare (e.g., Sarch 2015, 143–46). Moreover, the evidential account appears to have quite a limited domain of applicability (e.g., Hersch 2015, 282–83; also Hausman 2016, 29, acknowledging that the conditions posited by the evidential account, taken collectively, ‘are often not met’). These complications do not undermine the justifiability of relying on the evidential account in choice settings where the conditions posited by such account hold (e.g., Hausman 2022a, 355–56), but constrain the evidential account’s potential to ground reliable and informative welfare evaluations. In particular, they make it difficult to see on what basis philosophers/welfare economists may rely to establish whether the satisfaction of preferences provides reliable evidence for welfare in choice settings where philosophers/welfare economists are unable to determine whether the conditions posited by the evidential account are met and in choice settings where the conditions posited by the evidential account are not met (e.g., Fumagalli 2021, 126–28). In this respect, the evidential account’s purported agnosticism concerning the correct theory (or theories) of welfare appears to significantly constrain the evidential account’s potential to ground reliable and informative welfare evaluations. ¹⁹

6. Objection from conceptual ambiguity

The objection from conceptual ambiguity holds that informed PSTW do not withstand scrutiny because informed PSTW are premised on dissimilar (and often conflicting) conceptions of preferences. The objection proceeds as follows. In the economic and philosophy of science literatures, multiple conceptions of preferences have been advocated, which rest on dissimilar (and often conflicting) presuppositions regarding the relationship between preferences and choices (e.g., Thoma 2021b; Vredenburgh 2024, on the relationship between preferences and actual or hypothetical choices), the causal bases of preferences (e.g., Guala 2019; Ross 2011, on the issue of whether preferences are more plausibly regarded as mental attitudes, dispositions, or behavioural patterns), and the nature of preferences (e.g., Broome 1993; Hausman 2012, ch. 7–8, on the issue of whether preferences are more aptly characterized as judgements or feelings). However, substantiating informed PSTW requires the proponents of informed PSTW to specify which conceptions of preferences they endorse and put forward convincing reasons/evidence in favour of such conceptions. For the plausibility of informed PSTW critically depends on the merits of the conceptions of preferences on which informed PSTW are premised (e.g., Dietrich and List 2016b). Regrettably, the objection goes, the proponents of informed PSTW have hitherto failed to address these specification and justification challenges (e.g., Lecouteux 2022). ²⁰

This objection correctly notes that substantiating informed PSTW requires the proponents of informed PSTW to specify which conceptions of preferences they endorse and put forward convincing reasons/evidence in favour of such conceptions. Still, there are at least two reasons to doubt that the objection undermines informed PSTW. First, the information and consistency conditions presupposed by informed PSTW impose several constraints on admissible conceptions of preferences (e.g., Hausman 2011, 7–10, on how the consistency conditions presupposed by informed PSTW imply that preferences are inherently comparative). This does not per se enable the proponents of informed PSTW to univocally determine what conception of preferences philosophers/welfare economists should adopt in specific contexts (e.g., Mandler 2005, 255–56, on how the plausibility of various consistency conditions may itself vary depending on what conception of preferences one endorses). Still, it enables the proponents of informed PSTW to significantly narrow down the set of plausible conceptions of preferences (e.g., Cozic and Hill 2015, 297–99).

And second, the proponents of informed PSTW may justify their reliance on informed PSTW without having to specify and support a single general conception of preferences. For the merits of different conceptions of preferences are plausibly taken to depend on the theoretical and pragmatic presuppositions of the models and the policy applications where preferences figure (e.g., Angner 2018, 675–79). And different conceptions of preferences may be suitable for distinct modelling and policy purposes (e.g., Beck 2024, 1444–50; also Vredenburgh 2021). To be sure, theoretical terms such as ‘preference’ may have specific pre-theoretic connotations (e.g., Hausman 1998, 197–98; Mäki 1998, 306, on folk psychological conceptions of preferences). Yet, these pre-theoretic connotations do not determine the meaning of the theoretical notion of ‘preference’ figuring in informed PSTW (e.g., Ross 2012, 182–84). Hence, the proponents of informed PSTW may consistently acknowledge the existence of such pre-theoretic connotations and advocate distinct conceptions of preferences (e.g., Guala 2012, 137–39).

A critic of informed PSTW may object that the proponents of informed PSTW should ‘clarify the concept of preferences [they endorse] rather than leaving preferences to be defined implicitly by formal conditions and [by their] explanatory and predictive practices’ (Hausman 2024, 213, italics added). The idea is that although the proponents of informed PSTW ‘are free to reconceive of preferences in any way they wish [the proffered] reconceptualizations are not beyond criticism’ (224; also Sen 1973, 259). However, the proponents of informed PSTW can draw on several considerations to assess the comparative merits of different conceptions of preferences and support the specific conceptions they endorse. To illustrate this, consider the ongoing debate concerning the comparative merits of behaviourist conceptions of preferences, which regard preferences as indexes of choices (e.g., Gul and Pesendorfer 2008), and mentalist conceptions of preferences, which regard preferences as mental attitudes (e.g., Rubinstein and Salant 2008).

Behaviourist conceptions of preferences appear to be more general than mentalist conceptions of preferences since mentalist conceptions ‘limit the attribution of preferences to those with the requisite mental capacities’ (Hausman 2024, 223). In particular, adopting a behaviourist conception of preferences allegedly enables philosophers/welfare economists to ‘black-box […] the psychological processes that lead to choice’ and thereby avoid ‘controversial substantive commitments about psychological processes’ (Thoma 2021b, 165). Conversely, mentalist conceptions of preferences purportedly provide a more informative evidential basis to assess individuals’ welfare than behaviourist conceptions of preferences (e.g., Sumner 1996, ch. 5). In particular, adopting a mentalist conception of preferences may usefully constrain philosophers’/welfare economists’ attempts to reconstruct informed preferences (e.g., Hausman 2011, on how information concerning individuals’ beliefs can help philosophers/welfare economists determine how individuals conceive of the choice options they face). These observations do not determine whether, in general, philosophers/welfare economists should adopt behaviourist or mentalist conceptions of preferences. For what conceptions of preferences philosophers/welfare economists should adopt may plausibly depend on various contextual elements such as individuals’ cognitive/computational abilities (e.g., Okasha 2016) and whether philosophers/welfare economists aim to ground individual welfare evaluations or social welfare evaluations (e.g., Moscati 2021). Still, they nicely illustrate that the proponents of informed PSTW can draw on several considerations to assess the comparative merits of different conceptions of preferences and support the specific conceptions they endorse. ²¹

7. Objection from conceptual replacement

The objection from conceptual replacement holds that informed PSTW do not withstand scrutiny because grounding reliable and informative welfare evaluations requires philosophers/welfare economists to replace preference-based approaches with non-preference-based approaches. The objection proceeds as follows. To ground reliable and informative welfare evaluations, philosophers/welfare economists should distinguish between fundamental attitudes that ‘are the starting point of deliberation [and] shouldn’t be changed by the reasoning process’ and non-fundamental attitudes that ‘may be formed in deliberation [and] can be described as mistaken’ in light of the fundamental attitudes (e.g., Thoma 2021a, 355 and 361, on the putative contrast between ‘fundamental desires regarding features of the available options’ and less fundamental preferences). Abiding by this distinction, however, would require philosophers/welfare economists to regard preferences as the outcome of reasoning processes that involve more fundamental attitudes than preferences and thereby abandon preference-based approaches (and, therefore, informed PSTW; e.g., Rizzo 2025, 10, holding that ‘in back of preferences is desire [and] what is prudentially good for the individual is what she desires’). ²²

This objection correctly notes that non-preference-based approaches may enable philosophers/welfare economists to ground reliable and informative welfare evaluations. Still, there are at least two reasons to doubt that the objection undermines preference-based approaches (and, therefore, informed PSTW). First, philosophers/welfare economists can ground reliable and informative welfare evaluations by refining (rather than replacing) preference-based approaches. To illustrate this, consider again the challenges that apparent inconsistencies in individuals’ actual preferences pose to philosophers’/welfare economists’ attempts to ground reliable and informative welfare evaluations on information concerning individuals’ preferences. The proponents of preference-based approaches have addressed various such challenges by distinguishing between different sets of preferences (e.g., Hausman 2012, 36–37, for a preference-based approach distinguishing ‘basic preferences’, which are independent of individuals’ beliefs about ‘the character and consequences’ of the available options, and ‘non-basic preferences’, which take into account these beliefs and may influence basic preferences in light of such beliefs; also Rubinstein and Salant 2012, 375, for a preference-based approach distinguishing ‘observed preference orderings’, which vary as the result of cognitive processes, and ‘underlying preferences’, which purportedly ‘reflect [individuals’] welfare’). ²³

And second, replacing preference-based approaches with non-preference-based approaches does not per se enable philosophers/welfare economists to ground more reliable and informative welfare evaluations. For the proffered non-preference-based approaches face major conceptual and practical challenges and radically diverge on a number of foundational issues. To illustrate this, consider desire-based approaches. These approaches face major conceptual and practical challenges (e.g., Thoma 2021a, 360, on cases where desire-based welfare evaluations are indeterminate because the involved individuals’ putatively fundamental desires are vague) and radically diverge on a number of foundational issues, including what notions of desire should ground welfare evaluations (e.g., actual versus informed versus ideal desires), on what grounds philosophers/welfare economists should differentiate between more or less allegedly fundamental desires (e.g., the mere fact that a desire happens to be ‘the starting point of deliberation’ falls short of implying that such desire ‘shouldn’t be changed by the reasoning process’) and what exactly the connection between the posited desires and individuals’ welfare is (e.g., constitutive versus evidential connection). These divergences do not exclude the possibility that specific desire-based approaches may ground reliable and informative welfare evaluations, but cast doubt on the claim that philosophers/welfare economists should replace preference-based approaches with desire-based approaches.

A critic of informed PSTW may object that the highlighted divergences between the proffered non-preference-based approaches point to open problems in these approaches, but do not justify philosophers’/welfare economists’ reliance on preference-based approaches. In particular, the critic may maintain that neuro-psychological findings may enable philosophers/welfare economists to reduce preference-based approaches to non-preference-based approaches grounded on empirical findings concerning the neuro-psychological substrates of choices (e.g., Glimcher 2011, ch. 6–8). The idea is that neuro-psychological findings provide ‘a tool for measuring preferences neurobiologically’ (Levy and Glimcher 2012, 1027) and enable policy makers to ‘design [policies] that maximize welfare’ (Loewenstein and Haisley 2008, 238).

However, the great heterogeneity of the neuro-psychological substrates of welfare-enhancing choices casts doubt on the prospects of reductive non-preference-based approaches. For a given neuro-psychological process may contribute to generating choices having rather different welfare implications, and dissimilar sets of neuro-psychological processes may contribute to generating choices having similar welfare implications across choice settings (e.g., Ross 2014; Schulz 2024). Moreover, preference-based approaches frequently enable philosophers/welfare economists to ground reliable and informative welfare evaluations without having to draw on specific assumptions concerning neuro-psychological processes (e.g., Fumagalli 2019, on informed PSTW; also Section 3). These considerations do not exclude the possibility of grounding reliable and informative welfare evaluations on non-reductive non-preference-based approaches. However, together with the open problems faced by such approaches, they justify philosophers’/welfare economists’ reliance on preference-based approaches.

8. Conclusion

Let us take stock. In recent years, several prominent authors have argued that PSTW fail to provide a plausible theory of welfare. In this article, I have explicated and addressed the most influential objections put forward against specific versions of PSTW, namely informed PSTW. In particular, I have argued that although PSTW face significant conceptual and practical challenges, the critics of PSTW have hitherto failed to substantiate convincing objections against informed PSTW. This result does not exclude the possibility that additional objections may be put forward against informed PSTW. Still, as things stand, it demonstrates that philosophers/welfare economists are justified in relying on such versions of PSTW.

More generally, I take the considerations in this article to contribute to the ongoing cross-disciplinary debate about the plausibility of different theories of welfare in at least two respects of wide interest to philosophers/welfare economists. The first contribution concerns the conceptual and practical import of the objections put forward against specific theories of welfare. To illustrate this, consider again the objections put forward against informed PSTW. As argued in the previous sections, various objections share a tendency to misrepresent model-specific problems and particular contested cases as general conceptual and practical challenges to informed PSTW. This, however, by no means implies that the proffered objections are without merit. On the contrary, such objections provide valuable critical insights concerning philosophers’/welfare economists’ ability to reliably reconstruct welfare-relevant preferences in specific choice settings (e.g., Section 4 on the constraints imposed by widespread choice inconsistencies), the descriptive/normative adequacy of specific information and consistency conditions (e.g., Section 3 on transitivity), and the alleged need to supplement these conditions with further conditions on welfare-relevant preferences (e.g., Section 2 on moral considerations).

The second contribution concerns the need to heed cross-disciplinary differences when assessing the plausibility of different theories of welfare. To illustrate this, consider again philosophers’ and welfare economists’ respective contributions to the debate concerning informed PSTW. On the one hand, philosophers frequently engage in this debate at a higher level of abstraction than welfare economists and occasionally seem to overlook that welfare economists’ model specifications allow for more flexibility in the definition of preferences than many philosophers seek (e.g., Section 6). On the other hand, welfare economists often gloss over what many philosophers regard as significant normative/evaluative questions concerning individuals’ welfare and occasionally seem to overlook philosophically motivated reasons to doubt that the satisfaction of empirically elicited preferences constitutes welfare (e.g., Section 5). In this context, philosophers’ growing attention to welfare economists’ modelling practices and welfare economists’ deeper engagement with philosophers’ normative/evaluative discussions can greatly advance the ongoing cross-disciplinary debate about the plausibility of different theories of welfare.