1 Introduction

Welfare subjectivists typically hold that an object, event, state of affairs, etc., \(x\) is good for a subject \(S\) only if \(S\) takes a valuing attitude (of the right sort) towards \(x\) (Dorsey 2017b: 687). It is then a principal theoretical question for welfare subjectivists what amounts to “the right sort,” over which various subjectivist theories compete. The discussion in this paper is in this tradition, attempting to offer a promising candidate for “the right sort.”

I start by illustrating a dilemma for welfare subjectivists. On the one hand, various philosophers have argued that traditional subjectivist theories, like the desire-fulfillment theory, are too permissive to appropriately account for the well-being of typical mature human beings (Parfit 1986; Feldman 2004; Crisp 2006). According to their worry, it is implausibly permissive in that it allows for the well-being of someone to be determined by that person’s desire of any kind, including irrational, ill-considered, or base ones. This observation has led to considerable numbers of works that attempt to refine the desire-fulfillment theory of well-being.Footnote 1 One of the most notable examples is a theory that refers to idealized desire, i.e., one that would be held in an ideal circumstance (Railton 1986), typically including informed desire (see Sobel 1994and Rosati 1995 for discussion and criticism thereof). Alternative theories focus on life satisfaction (Sumner 1996; Suikkanen 2011) or value-fulfillment (Raibley 2010; Tiberius 2018).Footnote 2 Thus, here is one horn of the dilemma: A subjectivist theory of well-being must involve considerably sophisticated attitudes in order to address the well-being of typical mature human beings.Footnote 3

On the other hand, these sophisticated-attitude theories are too restrictive to appropriately account for the well-being of various welfare subjects including newborns, those with profound cognitive impairments, or non-human animals. The problem is that a refined view of well-being is cognitively too demanding for such welfare subjects, who lack the requisite cognitive capacity (Lin 2017). To illustrate, a life-satisfaction theory of well-being assumes that a welfare subject has the capacity for considering one’s (stages of) life, which newborns do not have. However, it sounds implausible that newborns are by definition less happy than typical mature human beingsFootnote 4 or that their well-being is undefined. We should admit that a life-satisfaction theory of well-being is inappropriate for some welfare subjects. The same problem applies to other sophisticated-attitude theories, mutatis mutandis. Thus, here is the other horn of the dilemma: A subjectivist theory of well-being must not involve overly sophisticated attitudes in order to address the well-being of all welfare subjects.

This observation might lead us to welfare variabilism, according to which it is not the case that a single theory of well-being applies to all welfare subjects. Welfare variabilists will see no dilemma here since there will be two (or more) theories of well-being, one of which involves sophisticated attitudes (suitable for typical mature human beings), and the others involve less sophisticated ones (suitable for other welfare subjects). Notably, it has been pointed out that welfare variabilism appears at least prima facie implausible in that “it is difficult […] to see what could explain why a theory of welfare might be true of some subjects but false of others” (Lin 2018: 324). However, instead of delving into the variabilist–invariabilist dispute further, this article examines the possibility of a single subjectivist view of well-being that applies to any welfare subject and simultaneously addresses the alleged dilemma.

In a nutshell, the task of this paper is to provide a subjectivist theory of well-being that appreciates our daily intuition about the well-being of various subjects, from typical mature human beings to non-human animals. Besides, rather than merely describing the diversity of welfare subjects and their well-being, this article aims to offer a principled theory to account for the entire spectrum of well-being. Such a theory must satisfy the following criteriaFootnote 5—it should (i) involve suitably sophisticated attitudes to account for the well-being of typical mature human beings; (ii) not involve overly sophisticated attitudes to account for that of other welfare subjects; and (iii) be invariabilist. In particular, this paper examines the following view:

Most-Sophisticated-Attitude View (MSA)

  1. (1)

    An objectFootnote 6\(x\) is good for a subject \(S\) in proportion to the strength of \(S\)’s pro-attitude towards \(x\) if, and only if, that pro-attitude is among \(S\)’s most sophisticated types.

  2. (2)

    An object \(x\) is bad for a subject \(S\) in proportion to the strength of \(S\)’s con-attitude towards \(x\) if, and only if, that con-attitude is among \(S\)’s most sophisticated types.

To have an intuitive grasp of the point of MSA, let us assume that the well-being of typical mature human beings should involve stable and well-considered pro-attitudes upon reflection rather than myopic, impulsive, and hasty ones. Suppose further that such a sophisticated attitude is beyond the capacity of human newborns. In this case, the well-being of typical mature human beings and that of newborns is assessed in different ways, and MSA accounts for how and why. On the one hand, the well-being of typical mature human beings might be assessed in terms of, say, authentic whole-life satisfaction—i.e., the life-satisfaction theorists of well-being are partly on the right track—when and because such a form of pro-attitude is the most sophisticated type of pro-attitude available to typical mature human beings. On the other hand, the well-being of newborns might be assessed in terms of pleasure when and because, again, it is the most sophisticated type of pro-attitude available to newborns. In short, MSA is an attitude-based theory of well-being that is sensitive to the differences in the capacity of welfare subjects. This capacity-sensitive nature is what distinguishes MSA from other subjectivist theories, enabling MSA to overcome the dilemma mentioned above.

Section 2 illustrates the basic ideas of MSA in more detail. Section 3 elaborates on MSA’s theoretical features. After distinguishing two forms of perfectionism, I argue that MSA is a theory of well-being that is substantially non-perfectionist and procedurally perfectionist. Section 4 introduces more moderate variations of MSA. Section 5 demonstrates the theoretical relationship between MSA and neighboring theories of well-being before addressing possible objections to MSA.

2 The Basic Idea of MSA

There are various types of pro-attitude. Typical cases of pro-attitudes towards \(x\) include (to name a few) being psychologically attached to it, being pleased with it, desiring it, judging it as valuable, being authentically satisfied with it, and regarding it as a component of one’s conception of the good.Footnote 7 In the same vein, there are various types of con-attitude. Let us use valenced attitudes as an umbrella term for pro-attitudes plus con-attitudes. That is, a pro-attitude is a shorthand for a positively valenced attitude, and a con-attitude is for a negatively valenced attitude. For simplicity, I will sometimes focus only on the positive side, but the argument is meant to apply to the negative side (i.e., negatively valenced attitude and ill-being), mutatis mutandis. Notably, each valenced attitude indicates a value-wise relationship, positive or negative (or neutral), between the attitude’s holder and its target. Valenced attitudes are contrasted on this point with other psychological states lacking evaluative relationships, like ordinary beliefs.Footnote 8

Philosophers often think that, intuitively speaking, there is a significant difference among the various types of pro-attitude in terms of sophistication. For instance, it is not unreasonable to see “regarding \(x\) as a component of one’s conception of the good” as more sophisticated than merely desiring \(x\). This kind of comparison, even a crude one, is a necessary presupposition of the oft-heard claim that (say) the well-being of typical mature human beings is assessed in terms of something more sophisticated than pleasure or desire. Hereafter, I assume that such a comparison based on sophistication is allowed and, more importantly, has been shared by philosophers, whether implicitly or explicitly,Footnote 9 with possible but negligible disagreement about what is more sophisticated than what.

What does such a ranking of sophistication consist in? It is beyond the purpose of this paper to offer a conclusive view on this matter because, again, my point here is simply to reformulate the sophistication-based comparison that is relevantly common in the philosophical discourse on well-being. With this caveat in mind, one suggestion is that a valenced attitude is sophisticated to the extent that it presupposes an advanced cognitive capacity of the relevant subject. For instance, the pro-attitude of “regarding something as a component of one’s conception of the good” presupposes the cognitive capacity for making a rational life plan (Rawls 1999: Sects. 63–64), while merely desiring something does not.Footnote 10

To clearly restate the comparison of sophistication in terms of cognitive demandingness, I stipulate several binary relations between valenced attitudes as follows. One type of valenced attitude, \({A}_{1}\), is at least as sophisticated as another type, \({A}_{2}\), if and only if having \({A}_{1}\) is at least as cognitively demanding as having \({A}_{2}\). \({A}_{1}\) is more sophisticated than \({A}_{2}\) if and only if (i) \({A}_{1}\) is at least as cognitively demanding (viz. at least as sophisticated) as \({A}_{2}\) and (ii) it is not the case that \({A}_{2}\) is at least as cognitively demanding (viz. at least as sophisticated) as \({A}_{1}\). One possible interpretation of the relation of cognitive demandingness goes as follows: \({A}_{1}\) is at least as cognitively demanding as \({A}_{2}\) when all who can have \({A}_{1}\) can also have \({A}_{2}\). When \({A}_{1}\) is at least as cognitively demanding as \({A}_{2}\) but not vice versa (i.e., all who can have \({A}_{1}\) can also have \({A}_{2}\), but someone who can have \({A}_{2}\) cannot have \({A}_{1}\)), \({A}_{1}\) is cognitively more demanding than \({A}_{2}\). Suppose that all who can judge something as valuable (e.g., typical mature human beings) can also be psychologically attached to something. Suppose further that someone who can be psychologically attached to something (e.g., newborns and non-human animals) cannot judge something as valuable. In this case, judging something as valuable is said to be cognitively more demanding—and hence more sophisticated—than being psychologically attached to something.Footnote 11

Finally, and most importantly, \({A}_{1}\) is \(S\)’s most sophisticated type if and only if \({A}_{1}\) is at least as sophisticated as anything in the set of types of valenced attitude that is possible for \(S\); that is, \({A}_{1}\) is \(S\)’s most sophisticated type if and only if (i) \({A}_{1}\) is in the set of types of valenced attitude that is possible for \(S\) and (ii) nothing more sophisticated than \({A}_{1}\) is in that set. For instance, the well-being of typical mature human beings can be determined in terms of their authentic satisfaction when (i) “being authentically satisfied with \(x\)” is in the set of types of valenced attitude that is possible for typical mature human beings, and (ii) anything more sophisticated than it is beyond human capacity. As my definition suggests, two or more types of valenced attitude can be among the most sophisticated for a given welfare subject. Suppose that “being authentically satisfied with \(x\)” and “regarding \(x\) as a component of one’s conception of the good” are equal in cognitive demandingness. If nothing more sophisticated than the two is possible for typical mature human beings, then both types of valenced attitude are the most sophisticated for them.

3 Theoretical Features

Concerning typical mature human beings, the implication of MSA is similar to that of many other sophisticated-attitude theories of well-being, such as those appealing to authentic whole-life satisfaction or value-fulfillment (Sumner 1996; Raibley 2010; Dorsey 2012; Tiberius 2018). It is a reasonable stipulation that the well-being of typical mature human beings requires a type of valenced attitude with stable and long-term plans rather than myopic, impulsive, and hasty ones. Another essential feature of their well-being may be a fair degree of autonomy.Footnote 12 In contrast, MSA has quite different implications regarding welfare subjects other than typical mature human beings. Let us assume that “being psychologically attached to \(x\)” is the most sophisticated type of pro-attitude of non-human animals.Footnote 13 In this case, MSA determines their well-being in terms of psychological attachment, without requiring, say, authentic whole-life satisfaction or value-fulfillment. That is because such “over-sophisticated” types of valenced attitude are beyond their cognitive capacity. The same applies to other welfare subjects as well, mutatis mutandis.

MSA offers a coherent explanation of how and why various welfare subjects can fare as well as typical mature human beings. (As a common practice in philosophical discussion of well-being, I simply assume the interpersonal comparability of well-being.) Let us suppose, contrary to MSA, that the well-being of any welfare subject was determined in terms of authentic whole-life satisfaction. The welfare level of those who lack the requisite cognitive capacity (such as newborns, those with a profound cognitive impairment, or non-human animals) would be necessarily lower than that of typical mature human beings, or even undefined. I find this upshot implausibly high-minded as an understanding of well-being, far deviating from our ordinary discourse about happiness and well-being.Footnote 14 There is nothing strange about thinking of their well-being and imagining that, say, a newborn is as well-off as a typical mature human being. We might even hold that, say, a newborn can be better off than a typical mature human being when both enjoy the same amount of pleasure in the same way. MSA is compatible with this intuition. If “being pleased with \(x\)” is the most sophisticated type of pro-attitude of a newborn, then pleasure contributes to the well-being of a newborn. On the other hand, it does not contribute to that of a typical mature human being because she is capable of something more sophisticated than feeling pleasure.

Here I hasten to mention a possible objection to MSA briefly. The last claim—pleasure is entirely irrelevant to the well-being of typical mature human beings because they are capable of more sophisticated types of valenced attitude—might sound extreme and thus implausible. Some might concede that typical human well-being is largely determined by sophisticated attitudes such as authentic satisfaction, yet maintaining that other “less sophisticated” types of valenced attitude should also be relevant, even to a lesser degree. I find this objection reasonable, and Sect. 4 addresses it in more detail.

At its face value, MSA might appear a variabilist theory of well-being in admitting that different types of valenced attitude are relevant to the well-being of different welfare subjects.Footnote 15 To illustrate, a typical interpretation of MSA goes as follows: value-fulfillment is relevant to the well-being of typical mature human beings, while the well-being of newborns is determined in terms of pleasure. This observation notwithstanding, MSA is an invariabilist theory of well-being because it provides a unified explanation of what type of valenced attitude is relevant to who’s well-being: according to MSA, for any welfare subject \(S\), the well-being of \(S\) is assessed in terms of \(S\)’s most sophisticated type of valenced attitude.

Notably, MSA is a perfectionist theory in a sense. Perfectionism about well-being, in a nutshell, holds that “human well-being consists in the development and exercise of one’s natural or essential capacities” (Dorsey 2010:  59). We should cautiously distinguish between two sorts of perfectionism. In its ordinary sense, a perfectionist theory of well-being indicates what particular object (e.g., courage and knowledge) is good for a welfare subject. Call it substantial perfectionism. Clearly, MSA is no more compatible with substantial perfectionism than, say, the desire-fulfillment theory of well-being is. According to the desire-fulfillment theory in its pure form, whether an object \(x\) is good for \(S\) is determined entirely by \(S\)’s desire. In the same vein, according to MSA, whether \(x\) is good for \(S\) is determined entirely by \(S\)’s most sophisticated type of valenced attitude. That said, MSA involves a different sort of perfectionism, which I call procedural perfectionism.Footnote 16 A procedurally perfectionist theory of well-being indicates not what particular object is good for a welfare subject, but what determines what particular object is good for a welfare subject. Put differently, procedural perfectionism holds that a good life consists in a certain sort of valuing and the appreciation of that value.Footnote 17 If a typical mature human being, \({S}_{1}\), judges something as valuable, perhaps it is good for \({S}_{1}\). However, that is not only because \({S}_{1}\) has a pro-attitude towards it but also because of the excellence that \({S}_{1}\)’s relevant pro-attitude exhibits. Otherwise, \({S}_{1}\)’s valenced attitude alone does not convey any prudential value to its target. For instance, if \({S}_{1}\) is merely pleased with another thing, it does not contribute to \({S}_{1}\)’s well-being. This observation does not imply that pleasure is always irrelevant to well-being. Suppose another welfare subject, \({S}_{2}\), whose cognitive capacity is limited compared to that of \({S}_{1}\), can feel pleasure but cannot have any more sophisticated valenced attitude. If \({S}_{2}\) is pleased with something, it may well contribute to \({S}_{2}\)’s well-being, given that the mode of valuing exhibits \({S}_{2}\)’s excellence.

One of the most prominent features of MSA is that it is substantially non-perfectionist (or, more strongly, substantially anti-perfectionist) and procedurally perfectionist.Footnote 18 This theoretical feature frees MSA from some typical shortcomings of welfare-subjectivist views. A common challenge against the desire-fulfillment theory appeals to its unduly permissive nature, purportedly indicating the need for refinement. One might have irrational, ill-considered, or base desires, which (according to the objection) are irrelevant to the well-being of typical mature human beings. Notably, as to the well-being of this segment of human beings, MSA may have a similar implication to other sophisticated-attitude theories of well-being, such as life-satisfaction theories and value-fulfillment theories, given that life-satisfaction and value-fulfillment are among the most sophisticated types of valenced attitudes available to them. As long as such sophisticated-attitude theories overcome the inappropriate-desire-based objections, MSA is immune to that line of criticism.

Suppose that an intelligent person, \({S}_{3}\), has a pro-attitude towards the activity of counting blades of grass (Rawls 1999, Sect. 65). Let us assume that the most sophisticated type of valenced attitude of \({S}_{3}\) includes “having something as a component of one’s rational plan of life.” If grass-counting is a component of \({S}_{3}\)’s rational plan of life, then that activity contributes to \({S}_{3}\)’s well-being even if that activity does not reflect anything objectively desirable or excellent. In contrast, we need not see grass-counting as contributing to \({S}_{3}\)’s well-being just because \({S}_{3}\) has some pro-attitude towards grass-counting—e.g., \({S}_{3}\) is pleased with this activity—without this activity being a component of \({S}_{3}\)’s rational plan of life.Footnote 19 However, it is not the case that “being pleased with something” (and some other types of valenced attitude, for that matter) is inherently irrelevant to well-being. Let me highlight the following point: according to MSA, the welfare-relevant type of valenced attitude is determined by the cognitive capacity of the welfare subject at issue. That is a non-negligible difference between MSA and other sophisticated-attitude theories. To see the difference more vividly, let us assume (deviating from the original thought experiment of Rawls) that \({S}_{4}\) is a human toddler, whose most sophisticated type of pro-attitude is “being pleased with something.” Suppose that \({S}_{4}\) is beginning to learn numbers and that \({S}_{4}\) is now interested in counting the number of anything. Suppose, finally, that \({S}_{4}\) is pleased with the activity of counting blades of grass. In this case, there is nothing wrong to think that the grass-counting activity contributes to \({S}_{4}\)’s well-being. That is because being pleased with something is the most sophisticated type of valenced attitude available to \({S}_{4}\).

A similar point applies to adaptation. From the perspective of MSA, we should determine \(S\)’s well-being in terms of \(S\)’s most sophisticated type of valenced attitude, regardless of whether they are adaptive or not. That is, it is not true that adaptation as such makes any difference to the assessment of well-being (see also Baber 2007). Suppose that a subject \(S\) has an adaptive pro-attitude towards an object \(x\), where the (adaptive) pro-attitude in question is among \(S\)’s most sophisticated types. This is not an unrealistic situation; indeed, our innocuous desires or life plans are more likely than not to be adaptive, at least partly (Bruckner 2009).Footnote 20 In such a case, it is implausible to insist that the pro-attitude is irrelevant to \(S\)’s well-being simply because the attitude is adaptive. On the other hand, if \(S\) has an adaptive pro-attitude towards \(x\) and the (adaptive) pro-attitude in question is not among \(S\)’s most sophisticated types, then \(x\) is unlikely to contribute to \(S\)’s well-being. This is what happens in typically troublesome cases of adaptive preferences. However, as is mentioned above, it is not the fact that the relevant pro-attitude is adaptive that makes it irrelevant to one’s well-being. In this situation, \(x\) (or the pro-attitude towards it) is irrelevant to \(S\)’s well-being because the pro-attitude in question is not among \(S\)’s most sophisticated types of valenced attitude.Footnote 21

4 Variations of MSA

Some might object as follows; MSA is implausibly high-minded in that it completely ignores less sophisticated types of valenced attitude. For the sake of argument, let us suppose that the life-satisfaction theorists of well-being are on the right track and that “being authentically satisfied with \(x\)” is the most sophisticated type of valenced attitude available to typical mature human beings. Even in that case, as the objection goes, it might be inappropriate to completely ignore less sophisticated ones such as desire, psychological attachment, and pleasure (Hawkins 2010; Hooker 2015: 19–20). In the same vein, even if desiring is the most sophisticated type of valenced attitude available to children, less sophisticated ones such as psychological attachment and pleasure may have at least partial relevance to their well-being.Footnote 22 As a response to this challenge, let us consider the following view.

Rank-Discounting Most-Sophisticated-Attitude View (Rank-Discounting MSA)

  1. (1)

    An object \(x\) is good for a subject \(S\) in proportion to the strength of \(S\)’s pro-attitude towards \(x\). Within the boundary of \(S\)’s capacity, the more sophisticated the pro-attitude in question is, the larger the goodness of \(x\) for \(S\) is, other things being equal.

  2. (2)

    An object \(x\) is bad for a subject \(S\) in proportion to the strength of \(S\)’s con-attitude towards \(x\). Within the boundary of \(S\)’s capacity, the more sophisticated the con-attitude in question is, the larger the badness of \(x\) for \(S\) is, other things being equal.

Like the original MSA, its rank-discounting version is motivated by procedural perfectionism. Indeed, assuming that perfection comes in degree so that different prudential value is ascribed to different levels of perfection (Bradford 2016 : 124–25), Rank-Discounting MSA is an ecumenical amendment of the original MSA. Again, procedural perfectionism ascribes perfectionist goodness to valuing attitudes and activities that exhibit one’s excellence. A natural corollary of this view is that, say, an item supported by one’s most excellent attitude is ceteris paribus better for her than one supported by her second most excellent attitude, which in turn is ceteris paribus better for her than one supported by her third most excellent attitude, and so forth.

Rank-Discounting MSA can be instantiated in various ways, depending on how to “discount” less sophisticated types of valenced attitude. As a typical illustration, consider a geometric view.Footnote 23 Let \(\left\{{A}_{S1},{A}_{S2},\dots ,{A}_{Sn}\right\}\) be the set of types of valenced attitude available to \(S\). Let \({A}_{S\left[i\right]}\) represent the \(i\)th item in the rearrangement of that set based on the binary relation “at least as sophisticated as.” That is, \({A}_{S\left[1\right]}\) is at least as sophisticated as \({A}_{S\left[2\right]}\), \({A}_{S\left[2\right]}\) is at least as sophisticated as \({A}_{S\left[3\right]}\), and so forth. Let \(W\left(x,S\right)\) denote the goodness of \(x\) for \(S\) and let \({A}_{S\left[i\right]}\left(x\right)\) denote the strength of \({A}_{S\left[i\right]}\) (i.e., \(S\)’s \(i\)th most sophisticated valenced attitude) towards \(x\). The basic idea of the geometric Rank-Discounting MSA is represented as follows,

$$W\left( {x,S} \right) = \alpha ^{{i - 1}} \cdot A_{{S\left[ i \right]}} \left( x \right),$$
(1)

where \(\alpha\) is a real number such that \(0<\alpha <1\). Notably, MSA is a special case of the geometric Rank-Discounting MSA as represented in \(\left(1\right)\), since \(W\left(x,S\right)={A}_{S\left[1\right]}\left(x\right)\) when \(i=1\). However, MSA and its Rank-Discounting counterpart have different implications concerning less sophisticated types of valenced attitude. While MSA does not count them at all, Rank-Discounting MSA admits some relevance of less sophisticated types of valenced attitude.

Perhaps one can have two or more distinct types of valenced attitude towards a single object simultaneously. Suppose that \({S}_{5}\) has a pro-attitude towards \(x\) in three ways (e.g., \({S}_{5}\) is authentically satisfied with \(x\), desires it, and is psychologically attached to it). Suppose, in contrast, that \({S}_{5}\) has a pro-attitude towards \({x}^{{\prime }}\) in only one way (e.g., \({S}_{5}\) is authentically satisfied with \({x}^{{\prime }}\) as much as \(x\), but \({S}_{5}\) neither desires it nor is psychologically attached to it). In this case, it seems natural to hold that \(x\) is better for \({S}_{5}\) than \({x}^{{\prime }}\) is. Hence, the following theory, although provisional, is not unreasonable when we have n types of valenced attitude in mind.

$$W\left( {x,S} \right) = \mathop \sum \limits_{{i = 1}}^\text{n} \left[ {\alpha ^{{i - 1}} \cdot A_{{S\left[ i \right]}} \left( x \right)} \right]$$
(2)

To see an intriguing implication of \(\left(2\right)\), suppose that \({S}_{6}\) has a pro-attitude (\({A}_{S1}\)) and a con-attitude (\({A}_{S2}\)) towards \(x\), simultaneously. Let us assume that \({A}_{S1}\) and \({A}_{S2}\) are equal in their strength and that \({A}_{S1}\) is more sophisticated than \({A}_{S2}\). For instance, imagine that a philosopher is authentically satisfied (upon reflection) with having her draft exposed to an honest and intense discussion, but she is psychologically frustrated to get her draft criticized. She is satisfied with her career as an academic philosopher, so she by no means denies that sincere criticism is good for her, all things considered, accepting the frustration as the legitimate cost to improve her draft. However, she also believes that it would be better for her if, unrealistically, she could bypass criticism (and frustration involved) and still improve her draft as productively as through criticism. I find such an ambivalent attitude not unreasonable and suggest that \(\left(2\right)\) accounts for the situation. To the extent that \({S}_{6}\) has a pro-attitude (\({A}_{S1}\)) towards \(x\), it is pro tanto good for \({S}_{6}\), whatever type the pro-attitude is. Likewise, to the extent that \({S}_{6}\) has a con-attitude (\({A}_{S2}\)) towards \(x\), it is pro tanto bad for \({S}_{6}\), again, whatever type the con-attitude is. It is a matter of comparison whether (and to what extent) \(x\) is good or bad, all things considered, for \({S}_{6}\). If \({A}_{S1}\) is more sophisticated than \({A}_{S2}\) as we have assumed, \(\left(2\right)\) will hold that \(x\) is more likely than not to be good for \({S}_{6}\), all things considered and other things being equal, while conceding that \(x\) is still pro tanto bad for \({S}_{6}\).Footnote 24

One might see the situation differently, however. When a philosopher is authentically satisfied upon reflection with receiving sincere criticism, is it bad for her, even pro tanto, to get her draft criticized? Rank-Discounting MSA replies affirmatively to this question. Some, in contrast, may answer negatively. More generally, they claim that a more sophisticated type of valenced attitude “trumps” the welfare-relevant power of the less sophisticated ones. A typical illustration goes as follows (cf. Hooker 2015: 19–20). Suppose that there are two relevantly comparable lives, \({L}_{1}\) and \({L}_{2}\), each of which involves some degree of value-fulfillment. Suppose next that \({L}_{1}\) involves less amount of pleasure than \({L}_{2}\). If the degree of value-fulfillment is the same between \({L}_{1}\) and \({L}_{2}\), then a subject can be better off in \({L}_{2}\) than in \({L}_{1}\). On the other hand, if \({L}_{1}\) involves more degree of value-fulfillment than \({L}_{2}\), then a subject is better off in \({L}_{1}\) than in \({L}_{2}\), no matter how much pleasure \({L}_{2}\) contains and no matter how small the marginal degree of value-fulfillment in \({L}_{1}\) is. The basic idea behind this view can be represented as follows.

Lexical Most-Sophisticated-Attitude View (Lexical MSA)

Assume a subject \(S\) can have only two types of valenced attitude, either \({A}_{S\left[i\right]}\) or \({A}_{S\left[j\right]}\) (\(i<j\)), towards objects. An object \({x}_{1}\) is at least as good for \(S\) as another object \({x}_{2}\) if, and only if,

  1. (a)

    \(A_{{S\left[ i \right]}} \left( {x_{1} } \right) > A_{{S\left[ i \right]}} \left( {x_{2} } \right)\), or

  2. (b)

    \(A_{{S\left[ i \right]}} \left( {x_{1} } \right) = A_{{S\left[ i \right]}} \left( {x_{2} } \right)\) and \(A_{{S\left[ j \right]}} \left( {x_{1} } \right) \ge A_{{S\left[ j \right]}} \left( {x_{2} } \right)\) 

Unlike the two variations discussed so far, the explanandum of Lexical MSA is the binary relation of “is at least as good for \(S\) as” instead of the unary attribute of “is good for \(S\).” For simplicity, let us set this complication aside and assume that the essence of MSA can be recast in terms of binary relations as follows: \({x}_{1}\) is at least as good for \(S\) as \({x}_{2}\) if, and only if, \({A}_{S\left[1\right]}\left({x}_{1}\right)\ge {A}_{S\left[1\right]}\left({x}_{2}\right)\). (Remember that \({A}_{S\left[i\right]}\) denotes \(S\)’s \(i\)th sophisticated type of valenced attitude and, in particular, \({A}_{S\left[1\right]}\) refers to \(S\)’s most sophisticated one.) Thus, Lexical MSA is equivalent to MSA in effect when \({A}_{S\left[1\right]}\left({x}_{1}\right)>{A}_{S\left[1\right]}\left({x}_{2}\right)\). The two diverge when \({A}_{S\left[1\right]}\left({x}_{1}\right)={A}_{S\left[1\right]}\left({x}_{2}\right)\); the original MSA finds \({x}_{1}\) and \({x}_{2}\) as equally good for \(S\) without referring to less sophisticated attitudes, whereas its lexical counterpart turns to the second and subsequent sophisticated ones until finding the “tie-breaker” between \({x}_{1}\) and \({x}_{2}\).

Lexical MSA is also backed up by procedural perfectionism discussed above, just like original and rank-discounting variations. It also shares the degree-sensitive nature with Rank-Discounting MSA, attributing different perfectionist goodness to varying levels of sophistication. However, Lexical MSA is different from Rank-Discounting MSA about the condition under which less sophisticated types of valenced attitude are relevant to one’s well-being. Suppose that \(S\) has a pro-attitude (\({A}_{S1}\)) and a con-attitude (\({A}_{S2}\)) towards \(x\), simultaneously, where the two attitudes are equally strong and that \({A}_{S1}\) is more sophisticated than \({A}_{S2}\). While \(x\) may well be good for \(S\), all things considered, Rank-Discounting MSA admits that it is pro tanto bad for \(S\) as long as \(S\) has a con-attitude towards \(x\).Footnote 25 In contrast, in Lexical MSA, more sophisticated types of valenced attitude have a lexical priority over the less sophisticated. Hence, \(x\) is not even pro tanto bad for \(S\) by virtue of \({A}_{S2}\), provided that \(x\) is the target of \(S\)’s more sophisticated types of pro-attitude (i.e., \({A}_{S1}\)).

I introduced Rank-Discount and Lexical versions as more moderate variations of the original MSA. Admittedly, MSA might be over-demanding and sometimes counterintuitive because of its sole focus on one’s most sophisticated type of valenced attitude. Indeed, I must confess that I am inclined to find the rank-discounting version more plausible than its original or lexical counterparts. For simplicity, however, the remainder of my discussion will focus on the most straightforward variation—the original MSA.

5 Neighboring Theories and Possible Critics

This section compares MSA with its neighboring theories of well-being. MSA combines several welfare-subjectivist components, so it possibly belongs to the class of subjective–subjective hybrid theories (Hawkins 2010; Woodard 2016: 169–70) or subjective list theories (Lin 2016; Heathwood 2021: 71).

“Subjective–subjective hybrid” (or “subjective list”) is the name of a class of theories of well-being, not that of a particular theory.Footnote 26 At least two specifications are in order. First, what subjectivist components are in question? This question is about the kind of pro- and con-attitudes that is relevant to well-being. Second, how should we “combine” them? To see the second question more vividly, let us assume that \({A}_{1}\) and \({A}_{2}\) are the two types of valenced attitude to be combined. A subjective–subjective conjunctive theory holds that \(x\) is good for \(S\) if and only if (i) \(S\) has \({A}_{1}\) towards \(x\) and (ii) \(S\) has \({A}_{2}\) towards \(x\). Alternatively, a subjective–subjective disjunctive theory holds that \(x\) is good for \(S\) if and only if (i) \(S\) has \({A}_{1}\) towards \(x\) or (ii) \(S\) has \({A}_{2}\) towards \(x\). Still alternatively, some might hold a more moderate view that \(x\) is good for \(S\) (i) to the extent that \(S\) has \({A}_{1}\) towards \(x\) and (ii) to the extent that \(S\) has \({A}_{2}\) towards \(x\).

Some might classify MSA as a special case of subjective–subjective hybrid theory, which I would hardly dispute. However, we should carefully distinguish MSA from other variations of such a hybrid, like conjunctive or disjunctive theories. The central feature of MSA is that it answers the two questions in terms of the cognitive capacity of each welfare subject. At the outset of this paper, we saw why one’s capacity is relevant; an overarching theory of well-being, when insensitive to the difference in one’s capacity, cannot apply to a broad range of welfare subjects with diverse levels of cognitive capacity. That is, a considerably sophisticated type of valenced attitude may be beyond the scope of newborns, while a less sophisticated type of valenced attitude may be too permissive to assess the well-being of typical mature human beings. This observation also applies to subjective–subjective hybrid theories. Suppose that \({A}_{1}\) is more sophisticated than \({A}_{2}\). Also suppose that \({A}_{1}\) is the most sophisticated type of valenced attitude available to typical mature human beings and \({A}_{2}\) is the most sophisticated one available to human newborns. Consider a subjective–subjective conjunctive theory, according to which an object \(x\) is good for a subject \(S\) if and only if \(S\) has both \({A}_{1}\) and \({A}_{2}\) towards \(x\). Such a theory is under-inclusive in that it requires a newborn to be able to have \({A}_{1}\) to be faring well. Next, consider a subjective–subjective disjunctive theory, according to which \(x\) is good for \(S\) if and only if \(S\) has either \({A}_{1}\) or \({A}_{2}\) towards \(x\). Such a theory is over-inclusive in that it allows as much relevance of \({A}_{2}\) to the well-being of a typical mature human being as that of \({A}_{1}\). I conjecture that the same problem can be raised against any subjective–subjective hybrid theories without any capacity-dependent feature, mutatis mutandis (see also Lin 2016; 2017: p. 373).

The point applies to a complicated variation of subjective–subjective hybrid as well. Jennifer Hawkins provides a “sketch” of an interesting class of subjective–subjective hybrid theories (Hawkins 2010), which could be restated as follows: there is a threshold welfare level, \(\theta\), such that \(x\) is good for \(S\) in proportion to either (i) the strength of \(S\)’s \({A}_{1}\) towards \(x\) when \(S\)’s level of well-being is above \(\theta\), or (ii) the strength of \(S\)’s \({A}_{2}\) towards \(x\) when \(S\)’s level of well-being is at or below \(\theta\), where \({A}_{1}\) is considerably sophisticated and \({A}_{2}\) is fairly primitive. I suggest that such a theory is under-inclusive; it is implausible to assess the well-being of a newborn by \({A}_{1}\), even if he is extremely happy. Moreover, if \(\theta\) denotes the upper limit of the well-being of newborns (without a capacity to have \({A}_{1}\)), that would be implausibly high-minded as an understanding of well-being. Similarly, such a theory is over-inclusive as well; it is inappropriate to assess the well-being of a typical mature human being by \({A}_{2}\), even if her level of well-being is below the relevant threshold.Footnote 27

Indeed, Eden Lin has it that this kind of problem shows the drawback of welfare-subjectivism in general as follows:

Nearly all extant subjectivist views face a dilemma: either they have implausible implications about the welfare of newborns, or they exclude newborns from their scope but cannot plausibly explain this exclusion. (Lin 2017: 373)

Note, however, that the range of the “extant subjectivist views” Lin has in mind is quite limited. He assumes just three possibilities; judgment-based, desire-based, and value-based theories (Lin 2017: 356).Footnote 28 All three types of valenced attitude are highly sophisticated, about which Lin is correct. The problem is that there are other types—in particular, less sophisticated types—of valenced attitude as well. Although newborns might be incapable of some types of valenced attitude like judgment, they can be pleased with or psychologically attached to something.Footnote 29 Welfare-subjectivism, with a sufficiently broad range of attitude-types in mind, need not “have implausible implications about the welfare of newborns.”

Some might claim that pleasure and psychological attachment are kinds of psychological state rather than attitude. Hence, goes the objection, it might be a misnomer to call them valenced attitude. In response, I am happy to use “valenced psychological state” instead of “valenced attitude” (see my note 7). It is largely a terminological issue. However, there are some points worth highlighting. Let us start with a natural assumption that well-being is such a kind of value as characterized by the expression of “value for \(S\).” It is important not to confuse it with “value from the viewpoint of \(S\).” Having a viewpoint presupposes a decent level of cognitive capacity, which means that something can be good/bad from the viewpoint of \(S\) only if \(S\) has a considerably advanced cognitive capacity. This observation may constitute a reason not to understand well-being as a kind of value characterized by “value from the viewpoint of \(S\),” given that subjects without the requisite capacity can fare well or badly.Footnote 30 Perhaps this is what Lin established in effect. The problem is, welfare-subjectivism is not the claim that \(S\)’s well-being (or something’s value for \(S\)) is assessed in terms of something’s value from the viewpoint of \(S\). The gist of welfare-subjectivism, as I understand, is that \(S\)’s well-being (or something’s value for \(S\)) fits or resonates with \(S\) in the right way (Dorsey 2017a: 198–99; 2017b: 686–87; cf. Railton 1986). Whether the fit or resonance presupposes the viewpoint of \(S\) is a substantial question. The answer can be negative when \(S\) is a welfare subject without the relevant cognitive capacity, where \(x\) positively resonates with \(S\) when \(S\) is, say, pleased with or psychologically attached to \(x\).Footnote 31

Still, even such a tolerant view of “valenced attitude” might not convince the sceptics like Lin. He also rejects a variabilist view of well-being, according to which the well-being of a newborn is a matter of pleasure and that of a typical mature human being is a matter of value-related beliefs. His rationale for this rejection is, quite roughly, as follows. Suppose \(S\) is high in terms of pleasure. If hedonism is true for \(S\)’s well-being, the level of \(S\)’s well-being is high. Suppose next that \(S\) newly acquires the cognitive capacity to have value-related beliefs, with \(S\)’s hedonic status unchanged. In that case, says Lin, it is implausible that the level of \(S\)’s well-being must drop to zero simply because \(S\) acquired the capacity (Lin 2017: 360).

In response, I suggest that biting some bullets is a plausible way out. To see this point, consider Charlie, a person with a profound cognitive impairment.Footnote 32 Let \({\text{Charlie}}_{n}\) denote Charlie at the time \({t}_{n}\). Suppose that, at \({t}_{1}\), Charlie’s most sophisticated types of valenced attitude include “being pleased with something.” In that case, we focus on what \({\text{Charlie}}_{1}\) is pleased with to assess his well-being. For instance, getting a slice of cake is good for \({\text{Charlie}}_{1}\) if he is pleased with it; being made fun of by his colleagues is bad for \({\text{Charlie}}_{1}\) if he is frustrated with it. Now suppose that Charlie underwent cognitive enhancement. As a result, Charlie’s cognitive capacity at \({t}_{2}\) is in no way inferior to that of typical mature human beings. Suppose that \({\text{Charlie}}_{2}\)’s most sophisticated types of valenced attitude include “being authentically satisfied with something.” In that case, we assess his well-being in terms of his authentic satisfaction rather than his hedonic status. Perhaps the well-being of \({\text{Charlie}}_{2}\) depends on whether he keeps a good relationship with his romantic partner or whether his academic paper is published in a prestigious journal, rather than whether he has a slice of cake. It is because the authentic satisfaction of \({\text{Charlie}}_{2}\) consists in a successful romantic relationship or academic achievement, while he is merely pleased with having a slice of cake.

If we resort to the Rank-Discounting version of MSA instead of the original one, the implication would be even more plausible. Arguably, it might be an implausibly strong claim that a slice of cake makes no contribution to \({\text{Charlie}}_{2}\)’s well-being; he may well like sweets. Rank-Discounting MSA does not deny this possibility, while holding that (a) other things being equal, the goodness of a slice of cake for \({\text{Charlie}}_{2}\) is smaller than the goodness of academic achievement for \({\text{Charlie}}_{2}\) is, and that (b) other things being equal, the goodness of a slice of cake for \({\text{Charlie}}_{2}\) is smaller than its goodness for \({\text{Charlie}}_{1}\).

Let us return to MSA in its original form. The relevant point mentioned above applies to when Charlie’s cognitive capacity is diminished as opposed to enhanced.Footnote 33 Suppose that the cognitive enhancement Charlie underwent has a substantial side effect, due to which the cognitive capacity of Charlie at \({t}_{3}\) (i.e., \({\text{Charlie}}_{3}\)) is at the same level as that of \({\text{Charlie}}_{1}\). In that case, we assess the well-being of \({\text{Charlie}}_{3}\) roughly in the same way as that of \({\text{Charlie}}_{1}\). Let us further suppose that the level of \({\text{Charlie}}_{4}\)’s cognitive capacity is even lower than that of \({\text{Charlie}}_{1}\) and \({\text{Charlie}}_{3}\). Suppose that the only possible type of valenced attitude of \({\text{Charlie}}_{4}\) is “being psychologically attached to something.” In that case, we will assess his well-being in terms of what he is psychologically attached to. If \({\text{Charlie}}_{4}\) can be as well-off as other Charlies with higher cognitive capacity (which I find plausible), we cannot assess the well-being of \({\text{Charlie}}_{4}\) and that of other Charlies appealing to the same type of valenced attitude.

In short, the level of \(S\)’s cognitive capacity is relevant to how to assess \(S\)’s well-being. To think otherwise is incompatible with the view that various subjects, whose levels of cognitive capacity vary, can be equally well off. This observation makes MSA (and, for that matter, its Rank-Discounting or Lexical variations) worthy of serious consideration. Interestingly, the gist of MSA discussed so far has a broader relevance than it may appear. Some might find it unnatural that a single theory of well-being applies to all Charlies and prefer to say that different theories apply to (say) \({\text{Charlie}}_{1}\) and \({\text{Charlie}}_{4}\). However, for such a welfare-variabilist description to be principled and well-founded, it needs to be supplemented by a further criterion to determine which theory of well-being applies to whom. This is where the capacity-sensitive view of well-being supported by procedural perfectionism kicks in.

6 Conclusion

This paper suggested and examined the most-sophisticated-attitude view of well-being (MSA) with its two variations. I find it one of the most plausible subjectivist theories of well-being, worth of serious consideration. The appeal of MSA is that it is sensitive to the fact that various welfare subjects have different levels of cognitive capacity, so it applies not only to typical mature human beings but also to other welfare subjects. Moreover, MSA is an invariabilist theory of well-being, providing a unified explanation of what type of valenced attitude is relevant to the well-being of a given welfare subject and why.

It is one thing that MSA is the best form of subjectivist theories of well-being; it is quite another that welfare-subjectivism is plausible at all. That said, “in arguing for subjectivism, the first step is to find the most plausible articulation of its main themes” (Dorsey 2012: 442). Thus, by illustrating MSA as a candidate theory of welfare-subjectivism, this paper effectively contributes to the argument for welfare-subjectivism itself.