A forgotten distinction in value theory

Abstract

The debate on final value has been so far understood as a debate over what sort of properties final value depends on. The debate’s reliance on mere dependence has, I argue, made it very difficult for conditionalists to put forward a coherent positive alternative to intrinsicalism. Talk of dependence is too coarse-grained and fails to distinguish between different ways in which value can metaphysically depend on other properties of the value bearer. To remedy this, I propose that we bring back a ‘forgotten’ distinction between two ways in which value can depend on other properties. We should distinguish those properties in virtue of which a value is had—the grounds of the value—from those on condition of which it is had—which following Dancy I call the enablers of the value. With this distinction in hand, I offer a clear re-statement of the two main conditionalist accounts of final value: non-instrumentalism and non-derivatism. When understood not as making claims about the properties on which final value depends but rather as making more specific ones about the properties that ground final value, these accounts are perfectly coherent.

1 Introduction

Since Christine Korsgaard (1983) suggested that we should leave space for a distinction between final value, or the value something has for its own sake, and intrinsic value, or the value that only depends on the thing’s intrinsic properties, the debate in the theory of value has very much focused on the nature of final value. There are two main sides to this debate. On one side, the intrinsicalists defend G. E. Moore’s position and argue that final value is always intrinsic, and that intrinsic value is always final. On the other side, conditionalists follow Korsgaard in thinking that there are extrinsic final values.

For the most part, the debate between intrinsicalists and conditionalists has been characterised in terms of opposing claims about the non-causal dependence of final value on intrinsic properties. Noah Lemos (2022, p. 1) has recently characterised the debate as follows: ‘intrinsicalism about final value holds that the final value of a thing depends solely on its intrinsic nature’, whereas ‘conditionalism about final value holds that the final value of a thing can depend on non-intrinsic or extrinsic factors’. A similar presentation of the debate can be found on the conditionalist side. For Jonas Olson, intrinsicalism defends ‘the idea that the final value of a thing depends, or supervenes, exclusively on properties intrinsic to that thing’ (2015, p. 46) while conditionalism rejects it (2004, p. 33).

In this paper, I want to focus on a distinction that has been forgotten in the discussion. This distinction is not, like Korsgaard’s, a distinction between types of value, but rather a distinction in the ways value can metaphysically, non-causally, depend on other properties of the valuable thing.¹ I want to suggest that we should distinguish between different roles that properties of the thing can play in making it valuable. By making claims about the properties on which value metaphysically depends, the debate has so far failed to make space for more fine-grained distinctions between different roles that these properties play in making something valuable. We should distinguish, I contend, between those properties in virtue of which something has value—i.e., those properties that ground the value—from those properties on condition of which it has value—that enable the value.²

Despite not being a distinction in value, it is a distinction that value theory should take more seriously than it has so far. Intrinsicalists have repeatedly complained that the debate on final value is not a fair game for them. The conditionalists, they protest, have mostly limited themselves to enumerating objects which they take to have final extrinsic value, but they have never—or very rarely—put forward a positive account of what they take final value to be. As Miles Tucker (2018, p. 132) has put it recently, ‘the conditionalist challenges [the intrinsicalist] to a fight in the dark: [the intrinsicalist] is asked to show the importance of intrinsic goodness over its competitor, but the concept of final value has barely been explained’.

I want to suggest that the conditionalist’s refusal to put forward a crisp statement of their position is due to their lacking the metaphysical resources to articulate their position. By making this distinction between the properties in virtue of which something has value and those on condition of which it does, I claim, we can make sense of the central tenets of the most prominent conditionalist positions. The two conditionalist theories I will focus on are non-instrumentalism, mainly defended by Shelly Kagan (1998) but also put forward by Rabinowicz and Rønnow-Rasmussen (2000) and originally by Korsgaard (1983), and non-derivatism, first properly presented by Jonas Olson (2004) but also recently ascribed to by Krister Bykvist (2015), Garrett Cullity (2017) and Rønnow-Rasmussen (2022). As I understand them, these two conditionalist positions make more fine-grained claims about those properties in virtue of which something can be finally valuable and so talk of non-causal dependence is too coarse-grained to capture their accounts of final value. Without the distinction between these different forms of non-causal dependence, the accounts to which the conditionalist’s examples hint seem puzzling at best and incoherent at worst.

Apart from a contribution to the theory of value, this paper also intends to show that there still are unexplored and yet fruitful applications of the distinction between grounding and other forms ofnon-causal dependence like enabling, disabling, intensifying or attenuating. Beyond Dancy’s (2004) ground-breaking work in the theory of reasons, others have, in recent years, pushed for a more widespread use of these distinctions. Shlomit Wygoda Cohen (2021), for example, has argued that putting grounding apart from enabling opens up space for an intermediate theory in the subjectivism/objectivism debate in metaethics. Ralf Bader has also put the distinction to good use in several domains within metaethics (2016, 2017), while others have extended its application beyond metaethics to discussions in general jurisprudence (Monti, 2022). If what I suggest in this essay is right (and I think it is), then there are still further interesting applications for these more fine-grained metaphysical distinctions and in areas within metaethics that have not been considered so far. This conclusion should, therefore, be of interest not only to those engaged in the debate on value but also to those who seek to motivate the distinction between these forms of non-causal metaphysical dependence.

The paper will be organised as follows. First, I introduce non-instrumentalism about final value to motivate Tucker’s worry that there might not be a coherent theory of final value behind the conditionalists’ examples. I, then, move on to distinguishing dependence, grounding and enabling. Once the distinction has been drawn, I argue that the two main conditionalist accounts—non-instrumentalism and non-derivatism—should be understood as making claims not about the properties on which final value can non-causally depend, but more fine-grained ones about those properties in virtue of which something can have final value. I close the essay with some needed taxonomical work and show how we should map out the terrain in the discussion around final value once the distinction between dependence, grounding and enabling has been introduced.

2 Non-instrumentalism and incoherence

Conditionalists, I said, have been criticised for failing to give a fully fleshed-out account of final value. They usually put forward counterexamples to the intrinsicalist identification of final value with intrinsic value but do not go on to present a competing account of final value of their own. This does not mean, however, that the conditionalist does not have a positive account in mind. For the most part, they do tend to gesture at a positive account through the examples of final value they give.

One of the ways in which the conditionalists have characterised final value is by contrasting it with instrumental value. This was, in fact, Korsgaard’s initial way of opening up space between intrinsic and final value. Korsgaard argued that we should keep two distinctions in goodness apart. ‘One is the distinction between things valued for their own sakes and things valued for the sake of something else’, by which she means the distinction ‘between ends and means, or final and instrumental goods’ (1983, p. 170, emphasis added). The other distinction is the one Moore had drawn ‘between intrinsically good things versus extrinsically good things’. Final value is, Korsgaard considers, properly contrasted with instrumental, rather than extrinsic, value.

Rabinowicz, Rønnow-Rasmussen, and Kagan are of a similar mind to Korsgaard. Rabinowicz and Rønnow-Rasmussen (2003, p. 215) ‘understand final value… in the standard way, by contrasting it with value as a means (instrumental value) and value as a part (contributive value)’. Kagan (1998, p. 278) identifies final value with ‘the value that an object has “as an end”’. He notes that ‘some objects are valued merely as means to other objects’ and so ‘possess “instrumental value”’. Yet, this instrumental chain cannot go on forever, Kagan (1998, p. 279) thinks, and so ‘eventually—or so the thought goes—we must reach objects that are valuable as “ends” or “for their own sake”.

Tucker’s complaint is perfectly understandable for neither Korsgaard, Kagan nor Rabinowicz and Rønnow-Rasmussen have given us much of a sense of what instrumental value is, and so of how we should understand final value by contrast.³ They do offer a collection of objects that they take to be finally valuable, including things like Princess Diana’s dress or a sports car. However, as we will see and as Tucker anticipated, the examples the non-instrumentalist offers do not seem to point to a coherent definition of non-instrumental final value.

One could, first, try to understand non-instrumental value as value that does not depend on any causal properties of the valuable object.⁴⁵ Final value, the idea would be, does not depend on what the effects of the objects are or what the object can help bring about. This would explain, for example, why a rare stamp has final value for the non-instrumentalist (Beardsley, 1965; Kagan, 1998). The value the stamp has in being rare, despite depending on externally relational properties, does not depend on a causal relation. The same goes for Rabinowicz and Rønnow-Rasmussen’s (2000, p. 41) example of Diana’s Dress. The dress is valuable because of a relational property, i.e. being Diana’s, but not because of what it brings about. We can then treat other notorious examples, like Lincoln’s Pen (Kagan, 1998), Rae Langton’s (2007) wedding ring or Napoleon’s hat, similarly. These objects are finally valuable because their value does not depend on their causal properties.

The problem with that suggestion is that it is too restrictive. Non-instrumentalists about final value are very adamant about the fact that the final value of something can depend on its causal relations. Kagan (1998, p. 281) explicitly says that ‘among the relational properties that are relevant to [final value] are the causal properties of an object’. Rabinowicz and Rønnow-Rasmussen (2000, p. 41) agree with Kagan, and Korsgaard (1983, p. 185) offers a series of now-well-known examples that seem to suggest as much. To illustrate his point, Kagan (1998, p. 284) discusses a sports car, ‘capable of driving at extraordinary speeds while still handling with ease’. He thinks it is intelligible to say that this car would have final value because of ‘its ability to perform at a particular speed’.

We can try to amend our initial account of non-instrumental final value to accommodate Kagan’s car example. One could think that instrumental value is value that depends not only on the causal properties of a thing but also on the value of what the thing can bring about. ‘When we judge that a thing is “good as a means”’, Moore for example thought, ‘we are making a judgement with regard to its causal relations: we judge both that it will have a particular kind of effect, and that the effect will be good in itself’ (Moore, 1903, p. 22, emphasis added). Final value could, then, be the value of a thing that does not depend on the value of what this thing can help bring about. After all, Kagan argues that one can value the sports car for its capacity to go fast and independently from whether one takes going fast to be of any value.⁶

Unfortunately, this will not do either. The non-instrumentalists offer plenty of examples of finally valuable things that, if we were to follow this definition, would fall on the instrumental side of the classification. Kagan talks, for example, about practical arts and skills having final value. He says that, despite the fact that we might value fine cooking as a means to some delicious gourmet dish, ‘it is not an uncommon view to hold that such abilities are [finally] valuable as well’. Kagan, however, notes that the final value of fine cooking depends not only on those excellences that constitute the skill, but also on the value of what they can bring about. ‘We do not typically value excellence in any skill whatsoever—however pointless and useless the skill’ (Kagan, 1998, p. 284). It is an important part of why fine cooking has value, Kagan goes on to say, that it is useful. ‘Were culinary expertise to somehow lose its instrumental value (if we no longer needed food, and if it no longer gave us pleasure), it would lose at least some (and perhaps all) of its [final] value as well’ (Kagan, 1998, p. 285).

Another example of an object that the non-instrumentalist takes to be finally valuable and the value of which depends on the value of its effects is Korsgaard’s mink coat. A mink coat, Korsgaard claims, ‘can be valued in the way we value things for their own sakes’ and yet ‘were it not for the ways in which human beings respond to cold, we would not care about them or ever think about them’ (Korsgaard, 1983, p. 185). The value of the mink coat, even if final, depends on the value that keeping ourselves warm has.

Therefore, final value can, according to non-instrumentalism, depend not only on the causal relations of the object but also on the value of what the object is causally related to. But the non-instrumentalist account of final value might now start to sound a little bit puzzling. The final—i.e., non-instrumental—value of a thing can have the same necessitation or dependence base as its instrumental value. This is what Kagan takes to be the upshot of his account of final value. ‘I am arguing’, he says, ‘that something may have [final] value—in part, or even in whole—because of its instrumental value’ (Kagan, 1998, p. 286, emphasis added). In typical cases, the instrumental value of a thing will not contribute to the thing’s final value and so it will be, in Kagan’s terminology, ‘mere instrumental value’. But, in other cases, ‘the object [will be] [finally] valuable because of, or by virtue of, its instrumental value’(Kagan, 1998, p. 288, emphasis in original).⁷

If instrumental and final values can non-causally depend on the exact same set of properties, it cannot be their dependence bases that put these two values apart. If we were to follow the literature and tried to individuate values in terms of their dependence bases, then we would be, in these cases, incapable of coherently explaining how is it that the thing is valuable i) on two counts and ii) in two different ways. We could not explain, that is, why (to use Kagan’s own example) fine cooking has both final and instrumental value rather than just ‘mere’ instrumental value. To make sense of non-instrumentalism about final value we need, therefore, to find a way to distinguish values that share dependence base. Absent such a possibility, non-instrumentalism about final value would indeed end up making incoherent ascriptions of value.⁸, ⁹

When we try to cash out the account of final value that underlies the non-instrumentalist’s examples in terms of non-causal dependence, incoherence looms. This is, I believe, the reason why non-instrumentalists have struggled to provide a positive account of final value that could rival the intrinsicalist proposal. It is not, however, that there is no coherent account behind the non-instrumentalist’s examples. Rather, it is that this account requires a careful articulation of more fine-grained relations of metaphysical dependence than the discussion on value has allowed for so far. To remedy this, I will now bring back a ‘forgotten’ distinction in the way value can depend on other features. With these fine-grained distinctions in hand, I then offer a statement of non-instrumentalism (§4) and non-derivatism (§5) that coherently captures all the cases these accounts want to respectively capture.

3 Dependence, grounding and enabling

Value is not a property that one can have ‘on its own’, as it were. If a thing is valuable, it is always valuable because it has some further properties that ‘make it’ valuable. So, value is in this sense a ‘dependent’ property. This dependence is usually expressed by saying that something has value because of certain characteristics that it has. This ‘because’ is not a causal ‘because’. When we say that the Gothic chapel is beautiful because of its fan vault we do not mean that the fan vault brings about as an effect the beauty of the chapel. These properties, rather, non-causally make it the case that the building has the value that it has.

It is very common nowadays to understand this sort of non-causal, yet metaphysical, dependence in terms of grounding.Despite some disagreements, grounding tends to be understood as an asymmetric, transitive and irreflexive relation.¹⁰ In different domains, questions of non-causal dependence have, therefore, now been translated using ‘grounding’. Philosophers talk about the ‘grounds’ of reasons, consciousness, beauty, truth or belief. The theory of value has, however, been a little bit more reluctant to follow this trend. In the discussion on value, and on final value in particular, theorists have been happy to simply rely on the notion of dependence or, when they do cash it out, they do it in terms of the (now somewhat démodé) supervenience relation. This is how, as we saw, Olson (2015) characterised the relevant dependence relation. But he is not alone. In his latest contribution to the debate, Lemos (2022) also uses ‘supervenience’ and ‘dependence’ interchangeably, and Toni Rønnow-Rasmussen (2022) equally relies on ‘supervenience’ in his latest book (see ft. 4 and pp. 8–18).

What I take to be significant, however, is not the fact that the literature on value has cashed out non-causal dependence in terms of supervenience rather than grounding. What is important is that, in understanding the debate in terms of what sort of properties final value non-causally depends on, it has failed to distinguish between the more fined-grained roles that properties on which the value of something depends can play in making it the case that the thing has value. Not all properties on which value non-causally depends play the same role in making the value-bearer valuable. Not distinguishing between different relations of non-causal dependence is a problem that remains irrespective of whether we cash out ‘non-causal dependence’ just in terms of grounding, supervenience, or some other notion.

Johnathan Dancy has been one of the pioneers in distinguishing different relations of non-causal dependence.¹¹ Perhaps the easiest way to motivate the different ways in which considerations can contribute to making it the case that a thing has a given property is by means of examples. Let us start, then, by taking an example adapted from Dancy’s (2004, p. 44) rich repertoire.

Failing to meet my friend at the restaurant is wrong because i) I promised him I was going to be there but also because ii) my promise was not extracted from me by duress or subterfuge, iii) I am able to jump on the bus and go, and because iv) what I promised to do is neither immoral nor otherwise unreasonable.

All i) to iv) are features on which my action’s wrongness non-causally depends—my action is wrong (at least partly) because of i)-iv). Yet, they do not all seem to be making the action wrong in the same way. That I promised my friend to be there is that in virtue of which my action is wrong in a way that the fact that I am able to do what I promised is not. The fact that I am able to go to the bus stop and take the bus without further inconvenience seems to be playing a ‘supporting’ role, as it were. If I was physically incapable of going to the restaurant or it was otherwise impossible for me to do what I promised, my not going would not be wrong. It would not be wrong because the fact that I could not possibly do what I promised would disable my not keeping the promise from making my action wrong. The same goes for the fact that the promise was not extracted under duress. This is not what properly makes the action wrong, but it rather enables my promise to make it wrong.

Even if all the features on which a property non-causally depends play some role in making it the case that something has the dependent property, they do not all play the same role. For simplicity, I will leave what Dancy calls ‘intensifiers’, understood as those features that strengthen the contribution of other properties, and ‘attenuators’, those that weaken it, aside, since what I say about the ground/enabler distinction also applies (mutatis mutandi) to them. So we should distinguish, then, between those features in virtue of which something has a property and those on condition of which it does.¹²

As I said, what I take to be essential to my contribution is not importing ‘grounding’ to the debate on value, but rather drawing the distinction between what I take to be different relations of non-causal dependence. This clarification notwithstanding, I will understand the features in virtue of which a property is had as those that constitute the ground of the property. When it comes to value, then, the grounds of the value are not all the features on which the value non-causally depends, but only those features that properly make the thing valuable—the ones that could be properly referred to as the ‘value-makers’. These are the features of a thing that, in Dancy’s (2004, p. 85) imagery, ‘give’ the thing its value.

Sometimes, as in the above examples, the ‘value-makers’ only conditionally ground the value of the thing.¹³ When this is the case, the features of the thing that constitute the ground are able to ground the value only when the relevant enablers are present, and relevant disablers are absent. When the grounding is conditional, the necessitation or dependence base of the value (V) will, then, be composed of the grounds (Γ) together with those other features that enable the grounding relation to obtain and the absence of those that would disable it (C), such that □[(Γ ∧ C) → V]. Features C are the features, we could say, on condition of which a thing has value. Therefore, whereas V non-causally depends both on Γ and C, it is only grounded in Γ (while enabled by C).¹⁴ Talk of mere ‘non-causal dependence’ obscures this distinction between the grounds (Γ) of V and the conditions (C), including enablers and the absence of disablers, of V.

Scepticism has probably been growing in your mind for a couple of pages now. So it might be time to address it. You might think that the distinction between grounds and conditions is not a ‘robust’ metaphysical distinction. You would not be alone. Roger Crisp (2000, pp. 36–40) has objected to Dancy’s use of the distinction in support of his reasons holism by arguing that the distinction only holds for non-ultimate grounds. Joseph Raz (2000, p. 67) had a different complaint. Raz argues that the distinction between grounds and conditions gets an ‘aura of plausibility’ because, when searching for explanations, some features are explanatorily salient while other features are taken as a given and so less salient. The distinction is, then, not a metaphysical one between different relations of non-causal dependence but a pragmatic one between salient and less salient features. This argument is usually supported by the observation that the parallel distinction between causes and background conditions is very probably pragmatic (Trogdon, 2013, ft. 30).

The first thing I want to note is that a defence of the robustness of the distinction is not something that I think is either possible in the limited space that I have here or required for the purpose of this paper. If, as I will argue below, the most prominent conditionalist accounts of final value should be understood using the distinction, then it is for the conditionalists to defend the distinction in order to defend their accounts. I am here only hoping to clarify what the tenets of conditionalism are, not to defend them.

With that note on scope in place, I still want to say a bit more to soothe sceptical worries. Even if it is true that Dancy has mostly defended the distinction by means of examples, and that these examples might be explained away as Raz and Crisp suggest, more substantive arguments have been put forward recently in favour of the distinction being a robust one. Wygoda Cohen (2020, p. 79) has, for example, noted that while grounding is (for most) an asymmetric relation, there appear to be cases of symmetric enabling. Bader (2016, p. 7) has also argued that attempts to collapse the distinction between those features in virtue of which a property is had and those on condition of which it is, like Raz suggests we do, cannot succeed. Not all the features that play a role as conditions of a grounding relation could play a grounding role. As we have seen in our example, some grounding relations are conditional on the absence of disablers and absences cannot play an active metaphysical role like grounding.

These considerations are good reasons for taking the distinction between grounds and conditions to be metaphysically robust. They also help, I think, in answering a related, but different, worry that a sceptic reader might have. One may suspect that the attribution of these fine-grained roles in particular cases will ultimately be guided by bare intuitions. Dancy (2004, p. 39) thought that there is not much more, in fact, than ‘intuitive training’ that would allow us to put conditions apart from grounds. But the sceptic will demand something more, especially if we want this distinction to do the work we want it to do in value theory and other areas.¹⁵ Luckily, I do not think intuitions are all we have got.

The arguments for the robustness of the distinction between grounds and conditions also give us non-arbitrary criteria that can guide our intuitions in the attribution of these fine-grained roles. If absences cannot be grounds, for example, we know that when the value of something depends on the absence of some feature, the absence should not be taken as part of the ground of the value. In this case, the absence plays a different role; it is a condition of the grounding relation. This is not the only consideration that can assist us in the attribution of these fine-grained roles. There is also a non-arbitrary way to reliably put enablers and grounds apart. Both enabling and disabling are, as Dancy (2018, p. 51) puts it, an ‘all-or-nothing’ matter. Enabling and disabling do not allow for degrees—they are ‘on’ and ‘off’ switches, as it were. An implication of this is that, when the dependent property allows for degrees (or is a determinable), it is the features properly grounding the dependent property that will determine how the grounded property is had—i.e. which determinate or to what degree it is had. As Bader (Unpublished, p. 7) puts it, the grounds, but not the conditions, stand in a functional relationship to the property being conditionally grounded. This will help us distinguish the grounds from the enablers of values, given that values do come in degrees.¹⁶

There is, then, a metaphysical distinction that the theory of value has, in relying on coarse-grained claims about dependence, forgotten. I say that the debate has forgotten, rather than plainly ignored, because the distinction had already been introduced to value theory by Dancy (2003, 2004) himself.¹⁷ Yet, saying that it has been forgotten might be, in a way, to put the point too lightly. Since Dancy introduced it, the distinction has hardly featured in the debate on final value. Most contributors seem to, in fact, ignore the distinction and those philosophers that do acknowledge it, such as Rabinowicz and Rønnow-Rasmussen, are quick to ‘disregard it as an interesting complication’ (2000, p. ft. 8). This has, unfortunately, not changed over the years. In recent work, Rønnow-Rasmussen (2015, 2022) puts once more the distinction aside doubting ‘that these details [about how to delineate value-making features from other kinds of features] would be significant in the discussions’ (2022, pp. 9–10).¹⁸ I, of course, think this is a mistake. The literature’s failure to distinguish the different relations of non-causal dependence that can hold between values and those properties on which value depends has made it very difficult for conditionalists to clearly state their positive accounts of final value. With the distinction in hand, then, let me move now to remedying this.

4 Non-instrumentalism in the light of the distinction

In §2, I said that non-instrumentalism seems to face a serious problem when we try to articulate the account of final value that informs the examples offered by non-instrumentalists. Non-instrumentalists argue that the final, non-instrumental, value of something can depend on the very same set of properties as its instrumental value. The problem is that, if we understand non-instrumentalism as a claim about what properties final value depends on, non-instrumentalism cannot distinguish a thing’s final, non-instrumental value, from its instrumental value when these two values share the same dependence base.

Pace Tucker, I do not think the problem is that non-instrumentalism is actually incoherent, but, rather, that talk of metaphysical dependence is too coarse-grained and so fails to pick up differences in the way in which value depends on its dependence base. As I argued in the previous section, there are at least two ways in which the value that something has can depend on the properties on which it depends. Some properties in the dependence base will be proper grounds of the value, while others will be playing the ‘supporting’ role of enablers or conditions. Sharing dependence base does not guarantee that one will also share grounds and conditions. Values V₁ and V₂ can both depend on the set of features {a, b}, but while V₁ is grounded in {a} and enabled by {b}, V₂ is grounded in {b} and enabled by {a}. So, two values with the same dependence base (i.e. {a, b}) could still be distinguished by the specific dependence relations that hold between the dependent value and the properties on which the value depends. It is only when there is complete identity not only in the dependence base but also in the roles that the features in the dependence base play that two values will be indistinguishable.¹⁹

We should, therefore, understand the non-instrumentalist account of final value not in terms of mere dependence but, rather, in terms of the more fine-grained grounding relation.

Non-instrumentalism: final value is the value of a thing that is not grounded in, even if it can depend on, the value of what the thing can help bring about.²⁰

The properties on which final value non-causally depends do not distinguish it from instrumental value. For non-instrumentalism, both final and instrumental value can depend on the value of what the value-bearer can bring about. Yet, while its instrumental value can and will be grounded in the value of what the thing can bring about, its final value cannot. So understood, there is nothing incoherent about non-instrumentalism nor about the examples with which it is generally motivated.

Kagan’s sports car has final value because its value is not grounded in the value of going fast, but rather in its capacity to go fast. The case of the mink coat—which could have appeared trickier at first—is not problematic at all. The mink coat has final value because of the value that being warm has for us. Yet, its final value is not grounded in that value, only enabled by it. It is not grounded in it because, as Korsgaard (1983, p. 185) notes, ‘people who want mink coats are not willing to exchange them for plastic parkas, if those are better protection against the elements’; but still its value is enabled by the value of its effects.²¹ If the conditions that make the coat have the valuable effect of keeping us warm were absent, the coat would lose its final value together with its instrumental one.

Something similar happens with fine cooking, the final value of which is not grounded in the value of delicious food but is still enabled by it. Once the distinction between grounds and enablers is introduced, we can easily explain, first, why there are two values and not just one and, second, why these two values are of different natures. The final and instrumental values of fine cooking, despite sharing dependence bases, do not have the same grounds and enablers. Fine cooking is, first, instrumentally valuable in virtue of the value of the food that the skills that constitute fine cooking can help bring about. However, fine cooking—or so argues Kagan—also has final value, in virtue of the skills that constitute fine cooking. The value of the food is relevant, however, but only in enabling these skills to ground the final value of the practice.

We have, then, a way of distinguishing final value from instrumental value, even in those cases in which a thing’s instrumental value depends on the very same properties as its final value. Instrumental value is not value that just depends, in any old way, on the value of what the thing can bring about, but, rather, is value that is grounded in the value of these effects. Final value is, by contrast, value that is not grounded in the value of what the thing can bring about—even if it can still be enabled by and so non-causally depend on this value.²²

Before I move on to the second conditionalist account of final value, let me note two things. First, one could still disagree with the substantive judgements that the non-instrumentalists make about the value of the mink coat, the car, fine cooking or Diana’s dress. Maybe we think that none of these things has any value or, at least, any final value. Yet, this is beside the point. The point is, rather, that the non-instrumentalists’ judgements about the value of these things are not at odds with each other and that there is a perfectly coherent account of final value underlying all these cases.²³ Second, I am not here arguing that the non-instrumentalists should revise their position to save it from incoherence. Rather, what I am saying is that the account was never incoherent to start with: it simply needed more careful articulation of two kinds of dependence relation.

5 Non-derivatism in the light of the distinction

Non-instrumentalism about final value is not the only conditionalist game in town. Some philosophers worry that non-instrumentalism is overinclusive. That a thing is not valuable for the sake of what it helps bring about does not imply, they think, that it is valuable for its own sake. Take Napoleon’s hat. Napoleon’s hat does not have value only for the sake of what it might bring about, sure enough. Still, one could think that it has value for the sake of something else, i.e. Napoleon’s sake.²⁴ Someone who despised Napoleon’s imperialist ambitions would hardly see any final value in his hat. The value of Napoleon’s hat, even if not instrumental, is still derived from the value of Napoleon’s historical feats.

So, final value should be contrasted—despite its non-instrumentalist sounding name—not with instrumental value but rather with derivative value. This is what the non-derivatists, as I will call them, propose. For non-derivatism about final value, 'the value of some F is final if and only if the value of F is such that it is non-contributory, i.e. if F is not (merely) valuable as a part of a valuable whole, and non-derivative, i.e. if the value of F is not derived from the (final or non-final) value of something else.' (Olson, 2004, p. 35, emphasis added).²⁵

If we treat contributory value as a species of derivative value (after all contributory value is value that the part derives from something other than itself), final value should be understood, according to the non-derivatist, as value that is not derived from either the final or non-final value of something else. A similar definition is given by Bykvist (2015, p. 134), who also understands final value as value that is ‘not derived from the value of anything else’.

The non-derivatist account of final value is more demanding than the non-instrumentalist account for it treats instrumental value as a species of derivative value. However, non-derivatism is still very much a conditionalist position. The value of something can be non-derivative and still extrinsic. Consider again the case of the rare stamp or Dancy’s (2004) example of the largest book in the Vatican.²⁶ The value that this book has for being the largest in the Vatican Library is surely extrinsic and yet the book is not valuable for the sake of something else. It is not valuable for the sake of something else, the non-derivatist wants to say, because its value is not derived from the value of something else to which the book is related—the smaller books need not have any value after all!

Despite offering some examples, the non-derivatist does not offer a straightforward characterisation of what ‘derivation’ is. Olson, anticipating a charge of incoherence from the intrinsicalist, is, nonetheless, quick to try to distinguish derivation from mere dependence. Olson argues that it would be a mistake to think that because the value of, say, Dancy’s Vatican book depends on the size of the other books, it is therefore derived from the size of the smaller books.²⁷ ‘To say that the final value of F depends on its context C is not to say’, he clarifies, ‘that the value of F is derived’ (Olson, 2004, p. 35, emphasis added). This is because ‘supervenience and derivation are different kinds of dependence relations’. Whereas supervenience ‘holds between the evaluative and the non-evaluative’, the relation of derivation ‘goes in the direction from the (final and non-final) value of something to the (non-final) value of something else’ (Olson, 2004, p. 35). To distinguish derivation from dependence (or supervenience in Olson’s terms), Olson gestures at the different nature of their relata. Derivation, Olson says, is a relation that holds specifically between values.

Olson’s way of drawing the distinction between derivation and supervenience will be good enough for cases like Dancy’s book or the rare stamp, where the value of F non-causally depends on externally relational properties of F but not on the value of those things to which F is related. Yet, it will fail to make sense of other cases that Olson and other non-derivatists would want to include as cases of final value. These are cases of values that the non-derivativist takes to be final but that depend on the value of something else. Olson (2004) and Cullity (2017), for example, discuss ‘appropriate pleasure’ as being something of final value, even if, we might think, its value depends on the value of its intentional object. One substantial complaint, raised by Tucker (2016), against non-derivatism is precisely that its definition of final value does not seem to cohere with the judgement that the non-derivatists make of these cases.²⁸ Tucker (2016, p. 1915) argues that, if we understand (as Olson does) non-derivative value as ‘value that does not depend upon the value of something else’, many of the well-known examples of final value given by non-derivatists would not have final value according to their own account.

If the non-derivatist wants to provide an account of final value that coheres with her ascriptions of final value, she needs to distinguish derivation from dependence. A mere distinction in the nature of the relata, as we have seen, will not do. It will not do because a relation of dependence between a dependent value and the value on which it depends does not necessarily give us derivation. There is something in the nature of the relation, and not just the relata, that puts derivation apart from mere dependence. Put differently, derivation is a way in which one value can depend on another value.

Derivation, I propose, should also be understood as a fine-grained dependence relation. Derivation is a form of grounding relation that holds between the derivative value and the one from which it is derived. Derivative value is not value that just depends, as Olson and Tucker suggest, on another value, but rather value that is grounded in the value of something else:

Non-derivatism: Final value is the value of a thing that is not grounded in, even if it can depend on, the value of something external to it.

As I argued in §3, we can have dependence without grounding. A value can depend on something else’s value without being grounded in that value. This is because, as I have been pointing out, there are other relations of non-causal dependence that can hold between two values and which are not grounding relations. A value can, for example, be enabled by the value of something else and so not be derivative as a consequence.²⁹, ³⁰

Understanding derivation as a more fine-grained relation fits better with the particular judgements that the non-derivatists make about certain cases. Korsgaard’s mink coat, for example, can be taken to be a case of non-derivative value. Its value depends on the value of its effects, but it is not grounded in that value.³¹ Korsgaard does not give us any precise sense of what could be grounding the coat’s value, but we could think that the mink coat has (for some) aesthetic value. The aesthetic value of the mink coat would be grounded, I imagine, in some characteristics of its fur, design and colours, but it could still be enabled by the fact that it has valuable effects. Uselessness could make these features look ridiculous or grotesque rather than beautiful.³² In There Are More Things, a horror story about a house not designed for human use, J. L. Borges put this thought brilliantly:

In order truly to see a thing, one must first understand it. An armchair implies the human body, its joints and members; scissors, the act of cutting […] None of the insensate forms I saw that night corresponded to the human figure or any conceivable use. They inspired horror and revulsion.

If Borges is right, then the mink coat’s final aesthetic value would depend on the value of its effects without being derived from it.

My account of derivation can also make sense of some examples of final value given by Olson himself. Like many in the literature, Olson focuses on the value of pleasures and pains. He thinks that pleasure can have final value but only when it is not directed at, for example, someone else’s pain. If John is pleased at Mary’s pain, Olson argues, John’s being pleased is not valuable. The value of John’s pleasure ‘(partly) supervenes on its intentional relation to Mary’s pain’ (Olson, 2004, p. 39). Note that it seems that what the value of pleasure non-causally depends on is the value of its intentional object. If Mary’s pain was an instance of ‘compassionate pain’, John’s pleasure would be valuable according to Olson.³³ So, the value of pleasure depends, it seems, on the value of something else. Yet, this does not preclude Olson (2004, p. 39) from thinking that John’s pleasure, when directed at the right object, is ‘valuable for its own sake’. How can this be so?

Once we distinguish derivation from mere dependence on value, it becomes quite clear that Olson’s treatment of John’s pleasure is perfectly coherent. The value of John’s pleasure depends on the value of Mary’s pain, yet it is not grounded in it. Dependence on value, as we have seen, is not enough for derivation. The fact that what John is taking pleasure in is valuable enables John’s pleasure to be valuable. In Olson’s (2004, p. 41) own words, ‘we rejoice in John’s being pleased on the condition that John is virtuous’. So, the value of John’s pleasure is enabled by and so depends on the value of that on which he takes pleasure without therefore deriving its value from it. That the value of John’s pleasure is not grounded in, and so does not derive from, the value of Mary’s pain can be seen from the fact that these two values do not have a functional relationship. An increase in the value of Mary’s pain would not result, all other things equal, in an increase in the value of John’s pleasure—only an increase or change in John’s pleasure would, on condition that Mary’s pain does not lose its value.³⁴

Dependence and derivation are, as Olson thought, different relations. However, they are not to be distinguished just by the type of properties that can figure as their respective relata. Derivation is a particular way in which value can depend on some other value. A value is derived from another when it is grounded in that other value.

Once more, I do not intend this account of derivation as a revisionary one. I consider, rather, that this is the relation that the non-derivatists have been trying to identify all along but could not because of a lack of the right metaphysical tools. Dependence is too coarse-grained a metaphysical relation to pick the derivation relation out. When the forgotten distinction between grounding and enabling is re-introduced, we can perfectly articulate non-derivatism about final value. When we understand derivation as a fine-grained grounding relation, it is clear that the non-derivatist treatment of cases like appropriate pleasure or the mink coat is compatible with and in support of the theory. Pace Tucker (2016, pp. 1918–1919), the fact that ‘the value of these things cannot be explained without appealing to the values of other things’ does not imply that these things ‘cannot have final value according to the [non-derivatist definition]’. As we have seen, some values, in being enabled by the values of other things, cannot be explained without appealing to other values but should still be taken to be final values according to non-derivatism.

Before moving on to the final taxonomical section, let me make one final point about derivation. Even though I have focused on the role derivation plays in non-derivatism about final value, it is worth noting that the concept of derivation is useful to and employed by intrinsicalists too. Many intrinsicalists are aware that final intrinsic value can still be derivative. Final intrinsic value cannot be derivative in the conditionalist sense of derivative because it cannot be derived from the value of something external to the value-bearer, but it can still be derived from the value of the value-bearer’s parts. Rønnow- Rasmussen (2015, p. 33) gives the example of a ring that is valuable only because of the diamond in its setting. In this case, we would say that the value of the ring, even if intrinsic, derives from the value of the diamond.

So the intrinsicalist thinks that there is an interesting category of value which they call ‘basic intrinsic value’.³⁵ Basic intrinsic value is value that is both intrinsic, so not dependent on externally relational properties of what is valuable, and non-derivative. Yet, the two main accounts of basic intrinsic value—put forward by Michael Zimmerman (2001) and by Timothy Perrine (2018, 2023)—face important shortcomings. As Lemos (2021) has recently argued, both accounts seem to be underinclusive. A full discussion of basic intrinsic value and the details of Zimmerman and Perrine’s accounts would require a full-length paper, but what I want to note here is that understanding derivation not just as non-causal dependence but as a more fined-grained grounding relation can also prove fruitful for an intrinsicalist account of basic intrinsic value.

Where I think both Zimmerman and Perrine go wrong is in taking the relation of ‘being a proper part’—or an ‘ontological ancestor’ in Zimmerman’s terminology—to be the only relevant relation when it comes to derivation. Yet, just having a part with basic intrinsic value is not enough for the value of the whole to be derived from it. The two values, that of the whole and that of the part, need to be related in a particular way for the former to be derived from the latter. The value of the whole needs, apart from depending, also to be grounded in the value of the part for it to be derivative and so non-basic. Relying on this more fine-grained notion of derivation would help the intrinsicalist accommodate the counterexamples that Lemos (2021) has put forward. For example, the state of affairs ‘At t, John knows that Carla courageously defends an innocent person’ can have basic intrinsic value because its value is not grounded in the value of the part ‘Carla courageously defends an innocent person’, even if this is a proper part of it and has basic intrinsic value of its own. Because what matters for derivation is grounding and not mere dependence, the same would be true if we considered the state of affairs ‘At t, John takes pleasure that Carla courageously defends an innocent person’. Even if we thought that, in this case, the value of the whole depends on the basic intrinsic value of the part ‘Carla defends an innocent person’, its value would not be derived from the value of this part because the value of the part ‘Carla defends an innocent person’ would be merely enabling the value of the whole.³⁶

The distinction between that in virtue of which something has value and that on condition of which something has it is neither a mere ‘interesting complication’ nor insignificant for the discussion. I hope that I have now given enough reasons to think that the debate on final value needs to be re-cast in terms of more fine-grained dependence relations. It is only when we understand the conditionalists, both non-instrumentalists and non-derivatists, as making claims not about mere non-causal dependence but more specific ones about grounding that we can articulate their positive accounts of final value and evaluate their merits properly.

6 Mapping out the debate

I want to, now, quickly run over how the debate would look like if we were to follow the suggestion of taking grounding, rather than the more general dependence, to be the relevant relation for conditionalists. In Table 1, I offer a brief taxonomical overlook of the main positions in the debate.

Table 1 Taxonomy of theories of final value

The positions diverge, then, on two dimensions and not just one. First, whereas intrinsicalists make stronger claims about dependence, conditionalists only make weaker claims about grounding. The second dimension along which these theories diverge is the one on which the debate has focused so far, that is the nature of the properties to which final value can be metaphysically related.

The strongest view is located in the bottom right corner and the weakest, more permissive, view is placed in the top left box of the table. Intrinsicalism is the strongest, more restrictive view, because it understands the final value of X to be value that is neither grounded nor dependent in any other way on the extrinsic properties of X. The weakest position is non-instrumentalism. Non-instrumentalism is the weakest because it allows for final value to non-causally depend on any type of property and only restricts those properties that can ground final value. This restriction is, also, weaker than the one imposed by non-derivatism. The non-instrumentalist does not think that being grounded in the value of just anything else is sufficient to make a value non-final, rather, it is only when the value of something is grounded in the value of what this thing can help bring about that its value is non-final.

The table also makes clear that the distinction between the grounding relation and the more coarse-grained dependence relation adds a whole new dimension on which different views about final value can diverge. It is only the Moorean intrinsicalist, in fact, who is committed to the strongest claim about non-causal dependence. Understanding the debate in terms of dependence has, therefore, done most of the views a huge disservice. Without the distinction between different ways in which value can non-causally depend on other properties, conditionalist views were forced to inhabit the right-side column and, consequently, to make claims that are way more restrictive than the ones they are willing to make. It is not a surprise, then, that many of the examples of final value that they offered were taken to fall without the boundaries of their own definitions of final value.

7 Conclusion

The intrinsicalists were right after all. The contest between intrinsicalism and conditionalism has not been a fair one. It has not been fair, however, because it has been set up in metaphysical terms that were better suited for intrinsicalism. For the conditionalist positions to have the metaphysical resources to be able to articulate their accounts of final value, we need to move away from claims of dependence and understand the debate in terms of the more fine-grained grounding and enabling relations. With these more fine-grained relations in hand, we can then offer, as I have done in sections §4 and §5, a clear and crisp statement of the central tenets of conditionalist rivals to intrinsicalism.

Let me close with some thoughts on what I take to be the most important implications, both for value theory and for metaphysical inquiry, of the foregoing discussion.

First, by putting forward a transparent rendering of the main conditionalist positions, this paper has made it clear that there are coherent and reasonable alternatives to Moorean intrinsicalism. Furthermore, it has also shown that conditionalism is not a view but rather a collection of, at the very least, two different views about final value. With their respective tenets distinguished, these accounts can now be evaluated on their own individual merits.

Second, by pointing out the relevant differences between dependence, grounding and enabling, this essay should also give an opportunity for the intrinsicalists to clarify their position. It is now the intrinsicalist who needs to make it clear what is the relation that she takes to be relevant for her claim. Until now, intrinsicalism has relied on claims about dependence but it is unclear whether this was the result of an active choice of metaphysical stance.

Third, as I argued in §5, understanding derivation as a type of grounding relation proves, in the theory of value, useful and avoids problems of overinclusion associated with more permissive accounts of derivation as mere dependence. I see no reason why this more fine-grained account of derivation as grounding should not deliver equally promising results in other domains.

Fourth, the role played by the distinction between grounding and enabling in cashing out derivation might lend further support for its robustness. The distinction between derivation and mere dependence on value is not a distinction that one can easily explain away as purely pragmatic. For one, it seems to have far-reaching implications for the relationship between rationality, motivation and value. As Korsgaard argues, if I only care about my mink coat for its instrumental value, I should be ready to replace it with better means of keeping myself warm. The grounding/enabling distinction, as I have argued, provides a straightforward explanation for why this is true of the coat’s instrumental value but not of its aesthetic value, even when its aesthetic value depends—as Borges might suggest–on the coat’s usefulness. Whether some other explanation for this divergence can be had is an important question but, until another one is found, the foregoing discussion seems to provide further good reasons to think that the grounding/enabling distinction is capturing something important.