2.1 Varieties of subjectivism

Before considering the problem that adaptive preferences raise for subjectivism, a few clarifications about subjectivism and the aim of this article are in order. As stated above, subjectivism contends that an object contributes to an individual's well-being if and only if this individual has a certain pro-attitude toward this object under certain conditions. Of course, several elements of this schematic formula need to be specified. Concerning the relevant pro-attitude, desiring is the most popular candidate. According to the desire-theory of well-being, an object contributes to an individual's well-being if and only if she desires this object under certain conditions. However, the desire-theory is only one of many potential options. For instance, one may claim instead that an object contributes to one's well-being to the extent that one values the object (Tiberius Reference Tiberius2018) or that one believes that the object is good for oneself (Dorsey Reference Dorsey2012). An alternative suggestion is that the relevant subjective pro-attitude is one of preferring an object to another (Baber Reference Baber2007). In this latter case, the subjectivist approach will not explicate the notion of well-being or what is good for an individual but of what is better for someone. In any case, the present article does not try to adjudicate these debates, although for simplicity's sake I will often present my discussion under the assumption that desire is the relevant pro-attitude. Moreover, I believe that subjectivists who disagree on that point will be able to transpose mutatis mutandis the argument that follows using their preferred pro-attitude.

I will also remain neutral on several other elements that a complete subjectivist account should specify. In particular, I will not decide whether the subject's well-being is increased when the object of the pro-attitude is instantiated or when the pro-attitude is held.Footnote ² Neither will I take a stance on whether what matters is the real fulfillment of desires or the belief that they are fulfilled.Footnote ³ Nor will I consider whether the good is the object of the pro-attitude, or the more complex state of fulfillment of the pro-attitude.Footnote ⁴

In short, the aim of my discussion is solely to assess whether the existence of adaptive pro-attitudes should lead subjectivists to introduce a condition of idealization or, on the contrary, to devise an appropriate actualism.

2.2 What is the problem of adaptive preferences?

Let us now turn to the problem raised by adaptive preferences or desires. To begin with, it would be useful to have a workable definition of adaptive desires. The general idea seems to be that a desire is adaptive if it is the result of a transformation or formation process that is prudentially beneficial by diminishing or avoiding frustration in one's circumstances. Yet, it is worth noticing that this definition allows for adaptive preferences that raise no problem for subjectivism. Bruckner (Reference Bruckner2009: 314) presents the case of Ursula, a woman who lost her husband. Before his death, she desired to spend her life with him. But presently this option is no longer feasible and, during the mourning process, she ends up losing her desire. This is surely an adaptation and it is moreover prudentially beneficial given her situation. Nevertheless, such adaptations don't raise any problem for actualism: to admit that her well-being after adaptation is no longer constituted by the desire which she has renounced does not seem either wrong or problematic. Although her situation is sad, it seems that we lack any ground to criticize her prudential adaptation.

In contrast, adaptive preferences that are prudential responses to oppression are obviously worrisome. Nussbaum describes the case of Jayamma, an Indian woman who works at a brick kiln and who carries bricks on her head all the day long without any hope of obtaining an easier, more valued or better-paid task. Moreover, although men start by also doing the same heavy labor, they rapidly graduate to the more skilled and less physically demanding tasks of brick molding and kiln loading, which are paid up to twice as much. Despite all this, Jayamma “seems to accept this as the way things are” (Nussbaum Reference Nussbaum2001: 23) and doesn't “think the division of labor at the brick kiln bad” (Nussbaum Reference Nussbaum2001: 140). According to Nussbaum, all this reveals that Jayamma has adapted her desires to her condition since this allows her to be satisfied with what she has and to avoid experiencing frustration.

If this is a real case of adaptive desires, a point to which I come back in the next paragraph, it seems to be a problematic one for an actualist version of subjectivism. Indeed, the result of her prudential adaptation is to diminish in virtue of unjust social circumstances the well-being that Jayamma can attain. The point is made salient if we consider a policy aiming at the welfare of Jayamma. If such a policy were to take Jayamma's welfare as consisting in the fulfillment of her adaptive desires, then it would unacceptably legitimize a distribution of labor that is obviously unjust.

Baber (Reference Baber2007) has however argued that there is little reason to believe that such examples really involve an adaptation of desires. Certainly, Jayamma accepts working at the brick kiln for only five rupees a day, yet it seems that this is merely because it is the best salary she can get. Why claim that she does not desire a higher salary? The fact that she does not hope to improve her salary merely reflects her beliefs about which desires can be satisfied. Jayamma surely adapts her choice to her beliefs about what is attainable, but there isn't any reason to suggest that she has also adapted her desires. Indeed, if a better employment with a better salary were offered to her, she would certainly not refuse it. She would not need to acquire or reacquire a desire for a better job. Thus, the possibility of an immediate reversal of her ‘adaptive desires’ seems to show that her desires have never changed. Hence, Baber claims that if adaptive preferences are ‘unstable’ in the terms of Elster (Reference Elster1983: 113), it should be clear that they are not adaptive: “if the fox is disposed to jump at the grapes as soon as they become available, then we should say that he has wanted them all along” (Baber Reference Baber2007: 112).

Now, does this show that the existence of problematic adaptive preferences is a myth and that they raise no problem for actualism? This would be too hasty because numerous adaptive preferences are not immediately reversible and often result from the internalization of oppressive cultural norms such as gender norms. For instance, people tend to see as most desirable the jobs that are considered suited to them within their culture. And once the preferences are built through these normative expectations, they are not easily reversed.Footnote ⁵ For a job previously unavailable and taken as inappropriate for women to become desirable, it is not sufficient that it becomes merely available. Notice, by the way, that this is why the real problem posed by adaptive preferences is not at its core a problem raised by a lack of information, or by momentary circumstances.

If all this is correct, then it appears that desires that are not immediately reversible and that are prudentially adapted to morally illegitimate circumstances are those that pose a real challenge to actualism. Certainly, such an adaptation to oppression is prudential and leads to a higher level of satisfaction given the oppression. But the problem is that the fulfillment of adaptive desires will often lead to a lower level of well-being than the fulfillment of the original non-adaptive ones would have. After all, gender norms aim to ensure that some attractive activities and positions are refused to women. Thus, the worry is that actualism may perpetuate the diminishment of well-being that results from people's adaptive preferences and therefore legitimize the initial injustice done to them.

I will therefore take it that adaptive preferences raise a real problem for subjectivism insofar as they are non-immediately reversible desires that have been formed as adaptations to morally criticizable circumstances.Footnote ⁶

2.3 Introducing an idealization condition to salvage subjectivism

Now that we have a better grasp on the adaptive preferences that raise a problem for subjectivism, let us see how subjectivism could overcome it. Since the problem targets the actual adaptive preferences of individuals, a natural solution is, at first sight, to introduce an idealization condition on preferences that will dispose of them. Such a condition may be either a restriction on actual preferences or a counterfactual condition that erases and replaces adaptive preferences. In the former case, I will talk of restricted actualism and in the latter of counterfactual subjectivism (or counterfactualism for short).

Whichever path we take, it remains to be settled which idealizing condition should be introduced. Insofar as adaptive preferences are adaptations to contexts of oppression, the introduction of an externalist condition of autonomy understood in causal or historical terms is surely an attractive option. If we introduce this condition as a restriction on the preferences that contribute to people's well-being, we have the following version of restricted actualism: a desired object contributes to a person's well-being if and only if this desire has been formed or could have been formed in a context free of oppression, manipulation, unfairness, and so on. A close but slightly different view would result if one uses Christman's theory of autonomy that offers an internalized version of such a historical condition. In his view, desires are autonomous to the extent that one would endorse their causal history from an unbiased, well-informed and rational standpoint (Christman Reference Christman1991: 11). We may then claim that a desired object contributes to well-being only if one would endorse the desire's causal history from an unbiased, well-informed and rational standpoint. In any case, whether we introduce a causal condition of autonomy or Christman's condition, they will both erase all adaptive preferences.

Alternatively, one may introduce an internalist condition of autonomy within a restricted actualism that will exclude only some adaptive preferences. Internalist versions of autonomy are numerous but they may all be understood as involving a form of endorsement or coherence. For instance, according to Dworkin's (Reference Dworkin1988) theory of autonomy, preferences are autonomous to the extent that they are or would be endorsed by their holders. Relying on this theory, Bruckner then claims that only preferences that are endorsed or would be endorsed under examination (Reference Bruckner2009: 322) contribute to their bearer's well-being. If one further adopts a hierarchical understanding of endorsement (Frankfurt Reference Frankfurt2018), one could then claim that desired objects contribute to people's well-being only if the latter desires are or would be after reflection the object of second order desires. Alternatively, one may prefer a stronger conception of endorsement according to which “[a]utonomy consists in deciding for oneself what is valuable and living one's life in accordance with that decision” (Colburn Reference Colburn2011: 61).Footnote ⁷ Building on such an account of autonomy, one would claim that only preferences one has decided to have or would have decided to have would make one better off. Finally, another avenue for internalist views about autonomy is to claim that autonomous preferences are those which are coherent.Footnote ⁸ For instance, one may suggest that preferences contribute to well-being to the extent that one would keep them under a condition of minimal revision for coherence (Dorsey Reference Dorsey, Räikkä and Varelius2013a). In any case, what is important with these internalist conditions of autonomy is that a restricted actualism that relies on one of them will not exclude all adaptive preferences since the latter may be endorsed or be part of a coherent set of preferences.

In sum, it seems as if an idealizing subjectivism has numerous options to overcome the problem of adaptive preferences. It can either introduce a restriction on preferences that excludes adaptive preferences or propose a counterfactual subjectivism that excludes them and replaces them. And it can do so for all adaptive preferences through an external/causal externalist condition of autonomy or only for a certain subset of them by relying on an internalist condition of autonomy or a mere condition of coherence.Footnote ⁹ The next section shows however that all these subjectivist conceptions of well-being that aim at excluding (some) adaptive preferences are defective in virtue of this very exclusion.

3.1 Against restricted actualism with an externalist autonomy clause

I first argue against a restricted actualism that relies on an externalist condition of autonomy by borrowing a case proposed by Khader (Reference Khader, Räikkä and Varelius2013). Consider Lucy, who is an elderly woman with a terminal disease. Being a full-time caregiver as her mother was, she believes that the role of women is to take care of others, especially of her husband and her children. She spent her life in accordance with this view, to which she fully adheres. In other words, she desired this life and did not desire anything else for herself. Moreover, she led a life that she sees as a happy one. Now, she is at the end of her life and afflicted with a terminal disease. Because she can no longer look after others, what she wants most is to get palliative care and die. Importantly, this is not because she suffers from her disease or is depressed. Rather, she feels that there is no sense in staying alive as a burden if she can no longer take care of others. In fact, she correctly anticipates that she would feel bad staying at home while being useless.

Obviously, her adherence to her lifestyle and more specifically her ensuing desire to get palliative care are adaptations of the worst sort since they are caused by social injustice. Therefore, her desire would be excluded by all versions of the historical clause: she did not control the conditions under which her desire was formed; the desire is the outcome of an unfair situation with regard to positions available to women. Even the weakest version of the historical clause proposed by Christman would exclude her desire: an unbiased Lucy would not endorse her desire if she was fully informed and aware that it resulted from oppression.

However, Khader claims that Lucy's actual strongest desire, which is to get palliative care and die, should be respected insofar as this is what will most contribute to her well-being in her present situation. I share this claim and I believe that there are several good reasons to accept it. The first is that she has a strong desire that is perfectly coherent with her life choices. Hence, to deny that fulfilling her desire will contribute to her well-being boils down to claiming that her extreme frustration, the impression that her will is not respected, will not downgrade her well-being. This seems hardly defensible. Of course, if there was enough time for her to become more autonomous, then one might think that discarding her present desire could lead to more well-being in the future. But this is precisely not her situation. Her frustration and the guilt she will experience in being, as she thinks, a burden for others, will not be compensated in any way. In addition, we know that becoming more autonomous often takes time and involves going through numerous hardships. Gaining more autonomy is not, at least not in Lucy's case as it is presented, merely giving up, all of a sudden, the desires and values around which her life has been constructed. Moreover, to claim that her most important desire should simply count for nothing in her well-being seems like an additional violation of her autonomy in the name of autonomy. Finally, it would even be cruel to destabilize her at the very moment at which she has to face her coming death, a moment at which we are certainly authorized to keep our innocent certitudes and desires, however mistaken they may be. I conclude that any restricted version of subjectivism that introduces a historical clause or any other clause the effect of which is to erase Lucy's adaptive desire is unacceptable.Footnote ¹⁰

3.2 Against restricted actualism with an internalist autonomy clause

In Lucy's case, a central consideration has been that her desire to have palliative care and die is endorsed or coheres with her other desires. This may lead one to suggest that a restricted actualism that relies on a condition of endorsement or coherence offers a better prospect. As we have already seen, such weaker forms of restricted actualism have been defended. Bruckner (Reference Bruckner2009) claims that as long as desires satisfy a condition of possible endorsement, their objects contribute to well-being. Similarly, Dorsey (Reference Dorsey, Räikkä and Varelius2013a: 45–46; Reference Dorsey2017: 206) argues that if adaptive preferences are deeply integrated and therefore cohere with one's other preferences, then there is no reason to reject them. Khader herself draws a similar conclusion from Lucy's case: the desires that should be taken into consideration, at least for decisions at the end of life, are those that would satisfy a condition of coherence.

Unfortunately, such a restricted actualism remains too strong a remedy for the problem of adaptive preferences. To see this, it suffices to slightly modify Lucy's case by introducing preferences that are not endorsed or lack coherence with her other desires. Suppose that Martha is perfectly similar to Lucy, except that she experiences from time to time dissatisfaction with her life. She sometimes dreams of another life, though her desire never led her to make real changes in her life. Or maybe she thinks that she would like to leave everything though she does not endorse this desire.Footnote ¹¹ However, like Lucy, Martha desires now to get palliative care and die.

Does the object of this desire contribute to Martha's well-being on a version of restricted actualism with an internalist autonomy or coherence condition? It seems that whatever the precise condition of that type we might consider, Martha's case can be designed to give a negative answer. For instance, one can stipulate that she didn't decide to desire to have palliative care and die; it only appeared to her as the best option. Or, she may not endorse her desire; she may have some weak second order desire to change her first order adaptive desires. Finally, her other first order desires may not be fully coherent with her desire to have palliative care and die. Hence, under this description, restricted actualism with a coherence condition will conclude that the object of Martha's incoherent or non-endorsed adaptive desire to have palliative care and die doesn't contribute to her well-being.

Yet, this response is unacceptable for the very same reasons that we have identified in Lucy's case. For sure, the tension in her preferences may leave room to help her to become more autonomous under some development of the case. But the crucial point is that it is not true for any development. Even if Martha's desires are incoherent or lack endorsement, it does not follow that her adaptive preferences contribute nothing to her well-being. Her actual more autonomous desires or her refusal to endorse her adaptive preferences may be too feeble for her to build her life on their fulfillment in the remaining time she has left. After all, her adaptive desires have been central to her whole life and the mere fact that Martha has an internal conflict may not diminish sufficiently the cost that being a burden for her family will imply for her given her present actual adaptive preference.

At this juncture, the defenders of a weak coherence condition of autonomy may reply that their view does not exclude Martha's desire (Dorsey Reference Dorsey, Räikkä and Varelius2013a). According to them, the fact that Martha's desire is in tension with some others is not sufficient to exclude it. For a desire to be excluded, it must be in tension with desires that are more numerous, weigh more or form a stronger web. If, on the contrary, as in Martha's case, the desire belongs to a set of desires we can counterfactually build by making her actual set of desires more coherent, then the desire contributes to Martha's well-being.

But this even weaker version of the coherence condition must also be rejected insofar as it still erases adaptive desires that contribute to well-being. Even if an adaptive desire is in tension with a weightier set of desires, it does not follow that it contributes nothing to well-being. First, to make such a claim, one would have to contend that the frustration of this adaptive desire will not diminish its holder's well-being. But nothing allows such a conclusion: if she decides to act in accordance with her most coherent set of desires and thereby frustrates her adaptive desire that is in tension with the latter, this will diminish her well-being. So, an assessment of her well-being cannot assume that the existence of her adaptive desire has no impact on her well-being.Footnote ¹² Second, if this person was transposed into a context in which her more coherent set of desires would be impossible to fulfill for various contingent reasons, then satisfying her adaptive desires may become what would contribute most to her well-being. But for a desire to contribute most to one's well-being in a certain context, its object must contribute to it in general. Therefore, since restricted subjectivism with an internalist clause excludes incoherent adaptive preferences, it cannot offer the proper account of her well-being in this circumstance.

I therefore conclude that even a restricted actualism that introduces something like an internalist autonomy or a coherence condition to exclude only some adaptive preferences should be rejected precisely because this exclusion remains a mistake.Footnote ¹³

3.3 Extending the argument against counterfactualism

The two previous sub-sections have shown that restricted actualism is mistaken. The exclusion of all or some adaptive preferences cannot be the right solution for subjectivism to overcome the problem raised by these preferences. If this is correct, then the argument immediately extends to a counterfactualism that appeals to an idealizing condition of autonomy or coherence. Recall that counterfactualism is the view that an object contributes to an individual's well-being insofar as he or she would desire the object under some idealized condition. Hence, if counterfactualism relies on an idealized condition of autonomy understood either in externalist or in internalist terms, then it will exclude respectively all or some adaptive preferences as contributing to well-being, and replace them with autonomous or at least coherent preferences. Since we have seen that this complete or partial exclusion of adapted preferences is unacceptable, counterfactualism falls prey to the same defects that have been pointed out in restricted actualism.

In fact, counterfactualism is even worse since it not only erases adaptive desires but also replaces them with counterfactual ones. Indeed, an immediate consequence of counterfactualism is that people's well-being can rely on desires that they actually lack. In other words, counterfactualism would then propose that some objects can contribute to people's well-being even if the latter are not motivated by these objects, a view of well-being that seems alienating (Railton Reference Railton2003). Since subjectivists have traditionally reproachedFootnote ¹⁴ objectivists for making this error, they surely should themselves avoid it.

In sum, it is not only restricted actualism but also counterfactualism that is unable to handle properly the problem of adaptive preferences. Moreover, insofar as their fault is to erase or replace adaptive preferences, any version of subjectivism that would erase or replace them must also be rejected. In other words, we have a good reason to adopt actualism without introducing any restriction on it.

3.4 Two further arguments for actualism

Beyond the arguments just presented, two general considerations also militate in favor of the contribution of adaptive preferences to well-being.

First, I believe that a general principle of practical reason vindicates that, even if it would be better to free oneself from some desires from a prudential perspective, they still contribute to one's well-being. Take for instance the desire to smoke. Even if it is rational to quit smoking now for its benefits in the long run, it might not be the best thing to do now if one is overloaded with work to finish or if one is confronting hardships, as long as there will be a more appropriate moment in the future and to the extent that the costs of delaying will be balanced by its benefits.Footnote ¹⁵ In other words, the following general principle of practical reason seems correct: even if a certain action constitutes a rational choice at a given moment, it may not be the best plan to follow if the cost could be reduced or more manageable in the future.Footnote ¹⁶ And from this general truth, it seems that a case can be made for adaptive preferences: even if it could be better to get rid of actual adaptive preferences for our benefit in the long run, the best plan could combine both their satisfaction in a short-term perspective and their elimination in a long-term perspective. Even if non-adaptive preferences will provide a higher level of well-being in the future, the best road for this self-transformation may encompass the fulfillment from time to time of adaptive preferences as long as one possesses them.

Second, we should note that adaptive preferences are rooted in desires that are important and are plausibly universal. For instance, fulfilling roles favored by culture is an important source of satisfaction because we thereby also fulfill desires to be respected, loved, well-integrated, valued for what we are doing, and so on. Hence, even if these roles put undue limits on one's long-term well-being, it does not follow that adaptive desires are merely superficial features of those who have them. The mere fact that it would be better not to have them does not mean that they could be erased as playing no role in well-being. On the contrary, they are often part of one's deep self and even when that is not the case, they cannot be counted for nothing.

The general conclusion is clear. Subjectivism about well-being should always take into account the objects of actual adaptive desires as contributing to well-being. This is not to say that the reasons to fulfill these desires will always outweigh the reasons to frustrate them. Surely not. This is only true in a minority of cases and especially in end of life contexts or situations in which people are under severe pressure. But for the objects of adaptive preferences to outweigh considerations aiming at the increase of autonomy in a minority of circumstances, one must never discard them from the outset. We first need to take the objects of people's actual adaptive preferences as contributing to their well-being in order to decide from there, depending on the particular context, which objects contribute most to their well-being.

4.1 Actualism at a time and over lifetime

If the argument so far is correct, actualism is right to count the satisfaction and frustration of one's present actual adaptive preferences as always contributing (respectively positively and negatively) to one's well-being. However, this is insufficient to claim that a subjectivist approach to well-being should remain actualist in response to the problems that adaptive preferences raise. After all, devising a welfare policy by relying on adaptive preferences without assessing them seems wrong because it seems that it unduly limits the level of well-being their bearers could attain. Moreover, we have noticed that if Lucy or Martha were not at the end of their life, it would be unacceptable to reduce their well-being to the fulfillment of their present actual desires, and among them their present actual adaptive desires. In Sen's words, this would “give sanctity” to the result of a process that is already unjust. We must therefore ask ourselves whether actualism is able to provide a satisfying treatment of this problem.

I believe that the most natural development of actualism can offer one. However, we must first develop actualism as a theory of well-being over a lifetime. To make a start, subjectivism must obviously contend that, beyond one's present preferences, one's future preferences are also relevant to one's well-being over a lifetime. In other words, a subjectivist theory of well-being over a lifetime must claim that well-being over a lifetime is constructed from the fulfillment and frustration of all the desires that one actually had, has and will have in one's life. In its simplest version, well-being over a lifetime is the aggregation of all the actual desires that one had, has or will have that were satisfied minus all those that remained unfulfilled.Footnote ¹⁷ Alternatively, one may defend more sophisticated rules of integration. In any case, there is no need to address this debate to show that an actualist subjectivism can overcome the problem of adaptive preferences. Therefore, I will stick here to the following general characterization of actualism over a lifetime:

A major reason to favor this general characterization is that it fits the temporal scheme that underlies nearly all conceptions of well-being over a lifetime. For instance, hedonism in its simplest form is generally understood as claiming that an individual's well-being is obtained by summing up her past, present and future pleasures and displeasures.

Now, an important consequence of TNA is that one may have prudential reasons to act in order to change one's desires whenever the fulfillment of the acquired desires would provide more well-being than our present ones and if these gains are not outweighed by the costs of this change. For instance, consider Arthur whose desire to play the piano as an amateur remains frustrating given, let us suppose, his lack of musical talent. Suppose now that once playing ping pong, he discovers that he has much more fun there. As a result, his desire to play piano is extinguished and is replaced by a desire to play ping pong. This is obviously a case in which someone is prudentially justified in changing his or her desires insofar as the change provides more well-being.

In sum, TNA implies that the consideration of one's good from a prudential perspective does not rely only on one's actual present desires but also on the potential desires one may actually acquire in the future and how they would contribute to one's well-being. As we will now see, this consequence of TNA provides a response to Sen and Nussbaum's objection.

4.2 How Temporally Neutral Actualism deals with adaptive preferences

Let us recall the objection. The worry is that the problematic cases of adaptive preferences, such as those which result from gender oppression, tend to diminish the well-being attainable by those who have these preferences. Hence, one may object that to adopt an actualist approach to subjectivism is just to accept this worse perspective for those who have adaptive preferences without offering the means to criticize and reverse them. And this, the objection goes, is just to legitimize the oppression they have suffered.

However, I will now argue that actualism understood as TNA does not have these implications. To start with, it is worth noticing that if most problematic adaptive preferences tend to lessen the level of well-being that will result from their satisfaction, this is not always the case.Footnote ¹⁹ For instance, suppose for the sake of the argument that the interdiction against smoking cannot be legitimate and that it is imposed in a certain country. Under these morally illegitimate circumstances, some of its citizens will adapt their desire; they may give up their desire to smoke and replace it with a healthier one. However, although the new desires are adapted, they could allow for higher levels of well-being. Therefore, when this is the case, Sen and Nussbaum's objection does not even get started insofar as one cannot claim that actualism diminishes the attainable well-being of those who have adapted their preferences.Footnote ²⁰

Therefore, for the objection to have some grip, we must consider problematic adaptive preferences the fulfillment of which diminishes the level of well-being that their holders may obtain in virtue of these adaptive preferences. Now, one might think that actualism faces an especially hard problem insofar as we have argued above that people's present adaptive preferences must always be taken into account as contributing to their well-being. However, and this is the crucial point introduced by TNA, it does not follow that these adaptive desires cannot change in the future for more autonomous ones. Moreover, the latter desires, if satisfied, would provide higher levels of well-being since we are precisely considering the cases in which adaptive preferences imply a diminishment of attainable well-being. Therefore, TNA does not imply that the well-being of people with adaptive preferences is definitely limited by their adaptive preferences. On the contrary, it underlines the possibility that in these circumstances, this diminishment may be reversed in the future, and that there is reason to reverse it grounded in an improvement of their future well-being. In other words, TNA allows one to see the negative effects of adaptive preferences on well-being and provides prudential reasons to overcome these negative effects.

To see this, consider a policy aiming at the well-being of citizens and, in particular, of those who have adaptive preferences. By definition, such a policy is concerned with the future well-being of these citizens. Now, if some preferences could lead to higher levels of well-being, then there are prudential reasons to acquire them. It follows that a policy aiming at well-being should rely on these prudential reasons. Hence, the policy is prima facie justified to help those who have adaptive preferences that diminish their attainable well-being to acquire more autonomous ones and to facilitate their fulfillment precisely in order to cancel this diminishment. In other words, even though some women under oppression have adapted their job ambitions to gender norms, a policy aiming at their well-being has prima facie reason grounded in their future good to help them, or at least allow them, to acquire and fulfill desires for the jobs that the gender norms refused to them. Thus, TNA is not limiting the well-being of people who have adaptive preferences to their fulfillment, pace Sen and Nussbaum. On the contrary, TNA prima facie justifies the promotion and fulfillment of non-adaptive desires that can be acquired and that will provide higher levels of well-being.

It is nevertheless important to emphasize two points about this response to Sen and Nussbaum. The first is that we are only explaining how subjectivism should take account of adaptive preferences. As such, TNA does not claim that some actions aiming at the well-being of people having adaptive preferences are pro tanto or even ultimately justified. Rather, TNA implies that people with adaptive preferences may have prudential reasons to change them for non-adaptive ones, from which it follows that policies or actions aiming at the good of the bearers of adaptive preferences may be prima facie justified. However, for this prima facie justification to become a pro tanto justification, it must also satisfy some autonomy or anti-paternalist constraints. Obviously, depending on one's favorite account of the latter constraints, very different policies will be pro tanto justified. One may either claim with Stoljar (Reference Stoljar, Veltman and Piper2014: 251) that some policies like incentive schemes will be pro tanto justified or on the contrary that this prima facie justification is always defeated by considerations of respect or autonomy.Footnote ²¹ In any case, TNA as such is neutral on this debate, and this is as it should be since a theory of well-being provides a basis for the justification of actions or policies aiming at people's well-being but certainly not their ultimate justification.

Second, it is important that this prima facie justification depends on the possibility of acquiring more autonomous preferences and on the gains that they would provide. As we have already seen, it may actually be impossible for some to change their preferences or to gain from this change. This in particular was our point in the case of Lucy. And even when a change is possible, it may be argued that the change is rarely or even never beneficial. For instance, Terlazzo (Reference Terlazzo, Vallier and Weber2017b: 245–246) argues that the change is never beneficial for adults, or at least for adults who are committed to their adaptive preferences. In any case, TNA in itself does not imply any such conclusions, since they will depend on psychological hypotheses that are independent of the TNA.

I therefore conclude that TNA provides a complete and appropriate treatment of adaptive preferences within a subjectivist approach of well-being. On the one hand, it insists that adaptive preferences are relevant to the well-being of their holder. It explains correctly that, under some circumstances, the most people with adaptive preferences could obtain is the fulfillment of these preferences. On the other hand, TNA makes full sense of the idea that while adapting their preferences has lowered their present attainable well-being, their well-being should not be unduly limited by them. This is because they may still acquire more autonomous ones in the future the fulfillment of which will provide higher levels of well-being. Hence, TNA does not legitimize oppression but offers the means to criticize its result and prima facie reasons to reverse the adaptive preferences.

4.3 More refined versions of idealizing subjectivism?

We are finally in a position to address a pending question: why not adopt a more refined version of subjectivism that would combine features of both actualism and counterfactualism? After all, our central argument against counterfactualism has been that it unduly excludes some adaptive preferences. A somewhat hybrid subjectivism could avoid this objection. I see two main avenues to construct a hybrid subjectivism.

The simplest strategy would be to claim that one's good is constituted by both actual and counterfactual desires. But this view does not fare better than the TNA. It has two defects. The first is that it would need to explain how one should compare from a prudential perspective actual and counterfactual desires. Should we give them equal worth? Or does the distance to a counterfactual desire diminish its contribution to well-being? And if the latter, how could we determine a decreasing ratio as a function of this distance? I do not see how we could offer justified answers to these crucial questions. In comparison, TNA avoids the problem. The accomplishment of possible desires makes one better off only to the extent that they can be possessed in the future and given the cost and probability of their acquisition and fulfillment. Notice that this response is not available to counterfactualism or mixed subjectivism, since the latter imply by definition that the fulfillment of a counterfactual desire that you may be unable to acquire nevertheless improves your well-being. If the desire cannot be acquired, you can have a measure neither of the probability of reaching it nor of the cost that this may involve. The second objection to this simple strategy is the one we have already pressed against counterfactualism: to claim that the fulfillment of a counterfactual desire one does not have and to which one is fully indifferent could make a difference in one's well-being is to offer an alienating conception of well-being.

An alternative strategy to combine features of actual and counterfactual subjectivism is Railton's conception of what is good for someone (Railton Reference Railton2003: 54). According to his proposal, an object is good for an individual if and only if this individual, ideally informed and rational, would desire that the individual she actually is desire this object. The strength of this approach is that it leaves open the possibility that the ideal individual would desire that her actual self desire precisely what this actual self desires. Hence, one cannot claim that it mistakenly denies that adaptive preferences contribute to well-being.Footnote ²² Nevertheless, one may wonder how the account deals with adaptive preferences. Unfortunately, no response can be drawn from Railton's proposal.Footnote ²³ But suppose that my ideal self would not desire that my actual self were to have my actual adaptive desires. Then, Railton's approach would fall prey to the objection we have raised against counterfactual subjectivism. So, for Railton's account to be correct, it must concur with actualism on its treatment of adaptive preference. Certainly, that may still leave room for a debate between TNA and Railton's view but it will have to rely on grounds that are external to the problem of adaptive preferences. Therefore, the point of the present article is preserved: insofar as adaptive preferences are concerned, subjectivism should be actualist.