1. The guise of the good thesis

I shall articulate the guise of the good thesis in terms of the correctness conditions of pro-attitudes. So articulated, it can avoid a number of standard objections. My formulation is inspired by G.E.M. Anscombe’s (Reference Anscombe1963) idea that:

On my proposal, just as truth is the correctness condition for belief, goodness is the correctness condition for pro-attitudes.

The guise of the good thesis is thus a thesis about pro-attitudes. Desiring (or wanting) something is a paradigm pro-attitude; pro-attitudes are sometimes characterized as being ‘desire-like.’Footnote ¹ Other pro-attitudes include those that are partially constituted by desiring, or by some ‘desire-like’ element or aspect, such as hoping, caring, and loving. Another pro-attitude will be important in what follows: preference. Just as desiring something is a pro-attitude towards that thing, preferring one thing to another is a pro-attitude towards the former thing.

Let’s begin with, as Anscombe puts it, the conceptual connection between ‘judgment’ and ‘truth, ’ or, as I shall say, the fact that truth is the correctness condition for belief. What this means is that:

This is the truth expressed by the slogan ‘truth is the aim of belief.’Footnote ² Note well: in (CB), ‘correct’ and ‘incorrect’ refer to representational correctness, i.e. the property had by representations when they are accurate or right, and representational incorrectness, i.e. that had when they are inaccurate, wrong, mistaken, or in error. Belief will serve as our paradigm case of something that can be correct or incorrect, in this sense. But there are other things that can be correct or incorrect, in this sense. At least some visual experiences, for example, are correct if and only if they are accurate or veridical, and visual experiences will emerge as an important case of thing that can be correct or incorrect, below.

It is important to distinguish the concept of representational correctness from two distinct, but perhaps related, concepts. First, the concept of representational correctness must be distinguished from the prescriptive concept of obligation (and its derivatives, prohibition and permission, as well as similar concepts of justification, virtuousness, rationality, and reasonableness). To say that it is correct (or incorrect) for someone to φ is not to say that they ought (or ought not) to φ. Consider the case of belief: it is correct to believe that p if and only if it is true that p, but this is not to say that you ought to believe that p if and only if it is true that p, since there are true propositions you ought not believe, such as those for which you have no evidence, and non-true propositions you ought to believe, such as those for which you have sufficient but misleading evidence. Second, the concept of representational correctness must be distinguished from the evaluative concepts of what is good, better, or best. To say that it is correct for someone to φ is not to say that it would be good, better (than some alternative), or best if they were to φ. Consider, again, the case of belief: it is correct to believe that p if and only if it is true that p, but this is not to say that it would be (for example) good if you were to believe that p, since there are true propositions such that believing them would not be good, such as the trivial truth that there are five paperclips on my desk right now, and non-true propositions such that believing them would be best, such as an athlete’s motivation-providing overestimation, in the midst of a match, of their chances of an unlikely, but still possible, comeback victory.Footnote ³

However, you might think that there are important connections between representational correctness and such prescriptive and evaluative concepts. Consider, again, the case of belief. Evidentialists in epistemology, for example, maintain that whether you are justified in believing that p depends only on the quality of your evidence relevant to the question of whether it is true that p. Reliabilists in epistemology, for another, maintain that whether you are justified in believing that p depends only on the likelihood that your belief is true given the way in which it was formed. (CB), you might think, is what makes these accounts attractive. However, such connections seem to hold in particular cases and not in general: visual experiences are correct if and only if they are accurate or veridical, but visual experiences do not seem to be the kinds of things to which prescriptive concepts apply – it is not, for example, that I have an obligation to have visual experiences that meet some criterion, determined by their correctness conditions.

Given these senses of ‘correct’ and ‘incorrect, ’ and following Anscombe’s suggested comparison, here is my formulation of the conceptual connection between ‘wanting’ and ‘good, ’ i.e. the guise of the good thesis:

And the same, mutatis mutandis, for other pro-attitudes (as well as for con-attitudes).Footnote ⁵ (GG) says that pro-attitudes are evaluative representations and therefore have evaluative correctness conditions.Footnote ⁶ Just as beliefs are correct if and only if their contents really are true, desires, for example, are correct if and only if their contents really are good. Just as beliefs are representations of truth, i.e. representations of how the world is, desires, for example, are representations of goodness, i.e. representations of how it would be good for the world to be.Footnote ⁷ However, note well that to say that beliefs are representations of truth is not to say that truth is part of the content of all beliefs – believing that p is distinct from believing that it is true that p – and that, likewise, to say that pro-attitudes are representations of goodness is not to say that goodness is part of the representational content of all pro-attitudes.Footnote ⁸

It will be important in what follows (Section 4) that (GG) is to be understood as an account of the correctness conditions of pro-attitudes, such that the defender of (GG) is committed not only to the truth of the articulated biconditionals, but also to the idea that, when it is correct to desire x, it is correct to desire x because, or in virtue of the fact that, x is good. Compare correct beliefs, which are correct because, or in virtue of the fact that, their contents are true.

I have followed Anscombe here in drawing an analogy between pro-attitudes and belief.Footnote ⁹ Some defenders of (GG), however, appeal to or suggest an analogy between pro-attitudes and visual experiences.Footnote ¹⁰ Their idea is that, just as visual experiences are representations of how things stand before the eyes of the experiencer, pro-attitudes are representations of the goodness of actual and possible states of affairs. Just as your visual experiences determine how things in your environment seem or appear visually to you, your pro-attitudes determine how actual and possible states and affairs seem or appear evaluatively to you. When you desire something, for example, that thing seems or appears good to you. Is (GG) consistent with the analogy between pro-attitudes and visual experiences? It is, given two points. First, seeming and appearance require representation: x cannot appear or seem F unless there is someone who makes some representation that is correct if and only if x really is F. Right now, my office door seems or appears brown, in virtue of my visually representing it as brown, but when I am napping later this afternoon, the door will no longer seem or appear brown, because I will no longer be visually representing it as brown. It is fine to understand (GG) in terms of how things seem or appear, so long as it is assumed that seeming and appearance require representation. Second, the analogy between pro-attitudes and visual experiences, even if it is illuminating, breaks down in one important way: in visual experience, when x appears or seems F to you, it is in virtue of your representing x as F, in the sense that F is part of the representational content of your visual experience, but in the case of pro-attitudes, when x appears or seems good to you, it is not in virtue of your representing x as good, in the sense that goodness is part of the representational content of your pro-attitude. As I mentioned above, (GG) does not entail that goodness is part of the representational content of all pro-attitudes. The sense in which pro-attitudes are representations of the goodness of actual and possible states of affairs is more perspicuously analogized to the sense in which visual experiences are representations of the truth about how things stand before the eyes of the experiencer. There is clearly a sense in which that is true, and that is the sense required to make the analogy between pro-attitudes and visual experiences illuminating.

Several standard objections to the guise of the good thesis target the idea that desiring something requires believing that it is good. (GG) does not have this consequence: it is consistent with (GG) that desiring x and believing that x is good are two different mental states, in the same way that a visual experience that represents x as F and a belief that it is true that x is F are two different mental states.Footnote ¹¹ Thus it is consistent with (GG) that it is possible to desire something whilst believing that it is not good and to believe that something is good without desiring it.Footnote ¹²

Why do its defenders accept (GG)? One basic motivation comes from the fact that it seems like it can explain what is wrong with certain intuitively problematic pro-attitudes. Consider that well-known passage in which Hume concludes that it is ‘not contrary to reason to prefer the destruction of the whole world to the scratching of my finger.’Footnote ¹³ Hume bases this conclusion on the premise that passions, including preferences, are not representations.Footnote ¹⁴ The defender of (GG) can diagnose his mistake: pro-attitudes are evaluative representations. Moreover, the defender of (GG) can explain what is wrong with Hume’s imagined preference: it is incorrect. It would not be better were the world to be destroyed (and my finger not scratched) than were my finger to be scratched (and the world not destroyed). Given (GG), therefore, it would be incorrect to prefer the former to the latter.Footnote ¹⁵ Hume sought to draw a distinction between beliefs, on the one hand, and pro-attitudes, on the other, by appeal to the idea that beliefs, but not pro-attitudes, can be correct or incorrect. (GG) rejects that account of the difference between beliefs and pro-attitudes: the difference between belief and desire is not that belief is representational and desire isn’t, but rather that belief and desire represent different things: beliefs are representations of truth, while desires are representations of goodness. (This provides an alternative account of the sense in which belief and desire differ in their ‘direction of fit.’)Footnote ¹⁶

2. The problem of partiality

(GG) says that pro-attitudes are evaluative representations, i.e. that goodness is the correctness condition for pro-attitudes (Section 1). This articulation, however, exposes a problem for (GG): the problem of partiality.

Like most people, I am partial towards myself and towards those to whom I am related by the bonds of friendship, on a broad sense of ‘friendship’ where this can include relationships between family members, colleagues, and compatriots: I prefer things going well for us to things going well for others. However, things going well for us is no better than things going well for others. Assume (GG). It follows that my partial preferences are incorrect. Why? Consider an individual case of partiality: we’ve both entered a raffle, and I prefer my winning the prize to your winning the prize. Assume (GG); it follows that my preference is correct if and only if my winning the prize is better than your winning the prize. However, my winning the prize is not better than your winning the prize – or so we can easily imagine: there is no relevant difference between us, nothing that makes me more deserving of the prize than you, no better consequences if I win and you do not, and so on. Therefore, my preference is incorrect. Generalizing from this, given some plausible assumptions about value, partial pro-attitudes are incorrect.

At first glance, this conclusion seems problematic. Partial pro-attitudes are a pervasive feature of normal human psychology; is it really true that they are all incorrect? Is there some way for the defender of (GG) to avoid this conclusion? That is the problem of partiality for the guise of the good thesis. I will argue, below (Section 6), that the defender of (GG) should accept the conclusion that partial pro-attitudes are incorrect. In the next three Sections (3 ⁴–5), I consider the prospects for defending the correctness of partial pro-attitudes, consistent with (GG) (Section 1).

3. Relative goodness

A natural thought at this point is that, in the raffle case (Section 2), although my winning the prize is not absolutely better than your winning the prize, it is better for me. Does this give the defender of (GG) a means of arguing that my preference is correct? (GG) says that pro-attitudes are evaluative representations (Section 1), i.e. representations of goodness. But do we mean absolute goodness, or relative goodness, or some combination of these? To take advantage of the idea that my winning the prize is better for me than your winning the prize, in defense of the correctness of partial pro-attitudes, the defender of (GG) will need to reformulate it so that it entails my preference is incorrect. Consider:

This seems to yield the desired result: it is correct for me to prefer my winning the prize to your winning the prize, because my winning the prize is better for me than your winning the prize.Footnote ¹⁷ However, (GG-egoistic) has some unhappy consequences. Consider a variant on Hume’s case (Section 1): imagine a far-away continent populated by millions of people whose destruction would have no effect on my wellbeing, and imagine that I prefer the destruction of this continent to the scratching of my finger, which would be (pro tanto) bad for me. In such a case, it would better for me were the far-away continent to be destroyed (and my finger not scratched) than were my finger to be scratched (and the far-away continent destroyed).Footnote ¹⁸ (GG-egoistic) thus entails that my preference, in this case, is correct. However, my preference, in this case, seems problematic in exactly the same way that Hume’s preference, in the original case, seemed problematic. Our amendment to (GG) has undermined one of the basic motivations for accepting it in the first place. It would be better to follow Hume in rejecting (GG), than to accept this amendment to it.

Let’s try a different approach. Perhaps the problem with (GG-egoistic) is that it focused too narrowly on what is better for the person who has the pro-attitude in question. Consider, then:

But this also has unhappy consequences.Footnote ¹⁹ Consider the arrival of a monstrous alien that benefits only from the destruction of all life on earth. Since the destruction of all life on earth is better for the alien than the preservation of some life on earth, it follows, given (GG-universal), that it is correct for me to prefer the destruction of all life on earth to the preservation of some life on earth. But that preference seems problematic in exactly the same way that Hume’s preference (Section 1) seemed problematic. Moreover, (GG-universal) entails that what seem like incoherent preferences can both be correct: assuming that the preservation of some life on earth is better for me than the destruction of all life on earth, it follows, given (GG-universal), that it is also correct for me to prefer the preservation of some life on earth to the destruction of all life on earth. It is thus correct both for me to prefer x to y and for me to prefer y to x. It would, again, be better to follow Hume in rejecting (GG), than to accept this amendment to it.

These problems will arise for any formulation of (GG) on which the mere relative betterness of x over y is sufficient for the correctness of preferring x to y. This idea is tempted when we think about what is good for human beings, as well as when we think of what is good for other living things – at least those more or less benign living things with which we are familiar. However, when we think of what is good for entities other than human beings, such as Microsoft, or of what makes for a good torturer or a good terrorist, this idea seems implausible.Footnote ²⁰ Moreover, relative goodness is cheap: as Judith Jarvis Thompson (Reference Thompson2008) notes, when speaking of relative goodness, ‘[e]verthing is good in some respect.’ (16; see also 10) But it is not, in virtue of this, correct to desire everything.

If the mere relative betterness of x over y is not sufficient for the correctness of preferring x to y, however, then the fact that my winning the prize is better for me than your winning the prize is not sufficient for the correctness of my preference in the raffle case (Section 2).

What is wrong, in the alien case, with preferring the destruction of all life on earth to the preservation of some life on earth? It is something like the fact that so preferring fails to take into account all that is relevant in this case. But all that is relevant … relevant to what question? The answer, it seems to me, is that so preferring fails to take into account all that is relevant to whether the destruction of all life on earth is absolutely better than the preservation of some life on earth, such as the fact that the destruction of all life on earth would be very bad for those creatures that live on earth. What we need to talk about, to make sense of what is wrong with such a preference, is what is absolutely better, and not just what is better for this or that person or creature, even if it turns out that the facts about what is absolutely better are determined by the facts about what is better for this or that person or creature. In any event, the defender of (GG), it seems to me, should maintain that preferences are representations of absolute betterness, and, in general, that pro-attitudes are representations of absolute goodness.

You might object that nothing is absolutely good. There are various species of relative goodness – for example, things can be good for me, in the sense of benefiting me, or, for another, an artifact like a knife might be a good knife, in the sense of being good as a knife – but, so the arguments go, there is no such thing as absolute goodness.Footnote ²¹ However, the thesis that nothing is absolutely good is consistent with my conclusion that pro-attitudes are representations of absolute goodness. If both of these claims are true, it follows that pro-attitudes are always incorrect. However, we should not automatically reject this consequence on the grounds that it implies pervasive incorrectness in our representations; such incorrectness is possible, as I will argue below (Section 6). What the defender of this objection will need to do is to argue both that nothing is absolutely good and that some pro-attitudes are correct. But, as I have been arguing in this section, it is unclear that there is any plausible way to formulate (GG) that has this consequence.Footnote ²²

In any event, we have failed to find a defense of the correctness of partial pro-attitudes (Section 2), consistent with (GG), that appeals to the fact that my winning the prize is better for me than your winning the prize.

4. Agent-relative goodness

In the previous Section (3) I expressed skepticism about formulating (GG) in terms of relative goodness, and in particular about formulating it in terms of the ‘good for’ and ‘better for’ relations. But there is a distinct kind of value relativity that philosophers have articulated, under the heading of ‘agent-relative value.’ Let us examine whether formulating (GG) in terms of ‘agent-relative goodness’ can help us defend the correctness of partial pro-attitudes.

The idea of agent-relative value, which contrasts with agent-neutral value, is derived from Thomas Nagel’s (Reference Nagel1970, Chapter X) discussion of agent-relative reasons (which he called at the time ‘subjective reasons), which contrast with agent-neutral reasons (which he called ‘objective reasons’). In Nagel’s paradigm case (90–91), if you are asked what reason you have for getting out of the way of an oncoming truck, you might cite either the fact that so doing will save your life – which is thus understood as an agent-relative reason – or the fact that so doing will save someone’s life – which is thus understood as an agent-neutral reason. Elsewhere (Reference Nagel1986; see also Nagel Reference Nagel1979; 101–104; Parfit Reference Parfit1984, 27), Nagel articulates the distinction as follows:

The idea is that whilst there is some reason that both you and I have to do something to get you out of the way of the truck – an agent-neutral reason – you, in addition, have a reason that I do not have – an agent-relative reason. Moreover, so the argument would have to go, this difference in reasons makes for a difference in value: your life, in this case, has both ‘agent-neutral value’ and ‘agent-relative value.’ Given that I have an agent-neutral reason to do something to get you out of the way of the truck, your life has some ‘agent-neutral value’ – your life is good, in an ‘agent-neutral’ way. However, given that you have an agent-relative reason to do something to get yourself out of the way of the truck, your life also has some ‘agent-relative value’ – your life is good, in an ‘agent-relative’ way, namely, in a way that is relative to you.

But isn’t this just to say that your survival is good for you, in the sense of ‘good for’ discussed in the previous Section (3)? This is suggested when reasons of self-interest are treated as the paradigm case of agent-relative reasonsFootnote ²³, but those who draw the distinction between agent-neutral and agent-relative reasons generally reject the conflation of agent-relative reasons and reasons of self-interest, given what work they want agent-relative reasons to do, e.g. to explain why I have a special obligation to take care of my family or why I have a special obligation to keep my promises. Footnote ²⁴ What is good relative to me – e.g. the welfare of my family or my keeping my promises – is not the same as what is good for me, in the sense of that which benefits me or is conducive to my well-being. I will reserve ‘goodness relative to’ for referring to ‘agent-relative value.’

Consider, then, the following reformulation of (GG):

Can we defend the correctness of partial pro-attitudes by appeal to (GG-relative)? This depends, first, on whether (GG-relative) implies that partial pro-attitudes are correct – e.g. on whether my winning the prize is better relative to me than you winning the prize (Section 2) – and, second, on whether (GG-relative) has implications that undermine one of the basic motivations for (GG) – e.g. on whether the destruction of the far away continent is better relative to me than the scratching of my finger (Section 3). How shall we answer these questions? We face an urgent question: what exactly does it mean to say that something is good relative to someone?Footnote ²⁵

Two ideas can immediately be rejected. First, we can reject the idea that something is good relative to someone if and only if they desire it. Combined with (GG-relative), this entails that, necessarily, all desires are correct. Second, we can reject the idea that something is good relative to someone if and only if it seems or appears good to them. The defender of (GG) finds attractive the idea that someone who desires x is someone to whom x seems or appears good (Section 1). But this thought, combined with (GG-relative), entails that, necessarily, all desires are correct. It would be better to abandon (GG) than to posit a species of representation that is necessarily correct.

So, again: what does it mean to say that something is good relative to someone? The difficulty here can be illustrated if we try to adapt Nagel’s definition of agent-neutral and agent-relative reasons. We cannot simply say that an agent-relative value is a value whose general form includes an essential reference to the person who has it – at least, not without giving an account of the obscure notion of someone’s ‘having’ a value. (We are familiar with the notion of someone’s having a reason.) Those defenders of agent-relative value who recognize this issue resolve it by appeal to some version of the ‘fitting attitude’ (or ‘dispositional’) account of value, and this is a common commitment among the diverse defenders of agent-relative value.Footnote ²⁶ Here is a basic version of such an account:

Given (FA), for every instance of goodness there is a corresponding fitting attitude. However, we may assume (in this context) that it is fitting for different people to have different pro-attitudes, and thus that the fittingness of pro-attitudes is agent-relative. We can thus understand agent-relative goodness as goodness whose corresponding fitting attitudes are agent-relative. This suggests the following definition of agent-relative goodness:

We thus have arrived at an understanding of agent-relative goodness. However, so I shall argue here, the defender of (GG) cannot endorse a ‘fitting attitude’ account of value.

Consider: what is the relationship between a ‘fitting’ attitude, as in (FA), and a ‘correct’ attitude, as in (GG)? It is hard to see how ‘fitting’ could here mean something different than ‘correct.’ ‘Fitting’ cannot, for example, mean ‘reasonable, ’ as there are situations in which it is reasonable for someone to desire something even though that thing is not good: if you have strong but misleading evidence that this glass contains gin (when in fact it contains petrol), it may be reasonable for you, assuming you would like to drink some gin, to desire to drink from this glass – but drinking from this glass is not good in any way. A ‘fitting’ attitude is just a representationally correct attitude (cf. Section 1), and ‘fitting’ and ‘correct, ’ as they appear in (FA) and (GG), mean the same thing,

At first glance, it now appears that (GG) and (FA) are not only compatible, but that they say the same thing. However, we must remember that (GG) is to be understood as an account of the correctness conditions of pro-attitudes, such that the defender of (GG) is committed to the idea that, when it is correct to desire x, it is correct to desire x because x is good (Section 1). Furthermore, as its name implies, the ‘fitting attitude’ account of value is to be understood as an account of value, such that the defender of (FA) is committed to the idea that, when x is good, x is good because it is fitting to desire x. However, if ‘fitting’ and ‘correct’ mean the same thing, then these respective commitments are incompatible – assuming that what explains x cannot also be explained by x – and so (GG) and (FA) cannot both be true. The same argument applies, mutatis mutandis, to (GG-relative) and (FA-relative). For this reason, the defender of (GG) cannot endorse a ‘fitting attitude’ account of value.Footnote ²⁸ I conclude that formulating (GG) in terms of agent-relative goodness cannot help us defend the correctness of partial pro-attitudes.

My argument in this section is based on the assumption that agent-relative goodness is to be explained by appeal to a ‘fitting attitude’ account of value. This leaves open the possibility of an alternative explanation of agent-relative goodness. In the absence of such an account, however, the defender of (GG) cannot appeal to agent-relative goodness in defense of the correctness of partial pro-attitudes.

5. Value and visual perspective

Recall the idea of an analogy between pro-attitudes and visual experiences (Section 1). Now consider the fact that our visual experiences are perspectival in at least two ways. First, our visual experiences give us a picture of our environment that is limited to those parts of our environment of which we are visually aware: we see only what is before our eyes, and not what is behind us or in some other place entirely. Second, our visual experiences give us a picture of our environment that is drawn, as it were, from a particular point of view: we see objects as being near or far, above or below, to the right or to the left, and so on. However, neither of these ways in which our visual experiences are perspectival provides for a way in which our visual experiences are incorrect – neither provides a way in which our visual experiences are inaccurate. My present visual experience, for example, is inaccurate neither in virtue of representing only what is before my eyes (as opposed to what is behind me, or how things stand right now at the Whale Pub in Hong Kong, or what was before my eyes yesterday) nor in virtue of representing what is before my eyes from a certain point of view (as opposed to from some other point of view, or perhaps sub specie aeternitatis). Here, then, is an idea: just as our visual experiences are perspectival, our pro-attitudes are perspectival – and the partiality of our pro-attitudes is just a symptom of the fact that they are perspectival.Footnote ²⁹ However, and most important, so the argument might go, just as our perspectival visual experiences are not thereby incorrect, our perspectival (i.e. partial) pro-attitudes are not thereby incorrect.

Let’s consider the two aforementioned ways in which our visual experiences are perspectival. First, so the argument might go, our pro-attitudes give us a picture of the evaluative landscape that is limited to those things that fall within the scope of our affective awareness, i.e. those things that are part of the content of our pro-attitudes. This seems like an apt description of some cases of partiality. Suppose some distant stranger has, unbeknownst to me, entered the raffle (Section 2). I do not plausibly prefer my winning the prize to their winning the prize; rather, I have no thoughts about them at all.Footnote ³⁰ They are analogous to some object of which I am not visually aware. Just as my visual experiences do not inaccurately represent such an object as non-existent, my pro-attitudes do not inaccurately represent my winning the raffle as better than the distant stranger’s winning the raffle. However, other cases of partiality seem different, including our paradigm case of the raffle (Section 2), in which I prefer my winning the prize to your winning the prize – you, whom I see hopefully clutching your ticket, wanting and deserving the prize no less than I do. It is not that you fall outside the scope of my affective awareness – you fall squarely within it, but I nevertheless prefer my winning the prize to your winning the prize. Partiality, in this and other such cases, is not merely a matter of our not having any attitude towards those outside our circle of friends. It is rather that we prefer things going well for us to things going well for them – i.e. they fall within the scope of our affective awareness. The analogy with visual perception thus breaks down in a crucial way.

Second, so the argument might go, our partial pro-attitudes give us a picture of the evaluative landscape drawn from a particular point of view, where point of view is understood not spatially but personally, in terms (among other things) of our interpersonal relationships with other people. The implications of this analogy with visual experience will depend on how we understand visual point of view, and there are two broad ways of understanding visual point of view, corresponding to representationalist and non-representationlist conceptions of perceptual experience. According to the representational understanding of visual point of view, visual point of view consists in your having visual experiences that represent spatial relations between you and the things that you see.Footnote ³¹ When you see two trees, one of which is closer to you than the other, for example, you visually represent the one tree as being closer to you than the other tree – and that is what it is for you to see the two trees from your point of view. However, there is a relevant disanalogy here between visual point of view and partiality. On the present proposal, visual point of view consists in the representation of spatial relations, e.g. the fact that (in the case described) the one tree is closer to you than the other tree. The analogue of these spatial relations, in the case of partiality, are interpersonal relationships, e.g. the fact that (in the raffle case) you are a stranger to me. Given (GG) (Section 1), however, our partial pro-attitudes do not represent interpersonal relationships. My preference in the raffle case, according to (GG), is correct if and only if my winning the prize is better than your winning the prize. There is a salient fact of interpersonal relationships here: my winning the prize is a winning of the prize by someone about whom I care a great deal and your winning the prize is a winning of the prize by someone about whom I care much less. But that fact is not what is represented by my preference, according to (GG). Given the representational understanding of visual point of view, therefore, visual point of view does not provide an apt analogy for partiality.

However, the analogy is no more apt when we move to the non-representational understanding. According to the non-representational understanding of visual point of view, visual point of view consists in your having visual experiences with certain qualitative, non-representational properties, which constitute what it’s like to have said experiences.Footnote ³² When you see the two trees described above, for example, the tree that is closer to you looks bigger than the other tree, despite the fact that you do not visually represent it as being bigger. However, according to (GG), partial pro-attitudes are evaluative representations. It is not just that we experience ourselves and our friends as being more important than others, in something like the way that we experience what is nearer as being larger than what is farther away. (GG) implies that we represent ourselves and our friends as being more important than others – this is what generates the problem of partiality (Section 2). Thus, given the non-representational understanding of visual point of view, visual point of view does not provide an apt analogy for partiality.

You might object that all this just shows that we were wrong to articulate (GG) in terms of representation (Section 1). Suppose that the nearer tree looks bigger than the tree that is farther away, although it is not represented as being bigger. Might we not say that, in an analogous way, that our friends are presented as being more important than others, although they are not represented as being more important? This would be implausible. The plausibility of the non-representational understanding of visual point of view derives from the premise that our visual experiences, in the relevant cases, have qualitative, non-representational properties – that there is something that it’s like to have them. The corresponding claim about partial pro-attitudes, however, is false: there is nothing that it’s like to prefer your winning the prize to my winning the prize. Preference does not require any distinctive phenomenology: it is possible to prefer one thing to another without their being anything that it’s like for you to prefer the one to the other. Seeing, you might think, is different: seeing requires that there be something that it’s like for you to see. Pro-attitudes do not essentially have qualitative, non-representational properties – whereas you might think that visual experiences do – and so, for this reason, we cannot articulate (GG) in non-representational terms.

6. Partiality as illusion

The problem of partiality (Section 3) is that (GG) seems to entail, along with some plausible assumptions about value, that partial pro-attitudes are incorrect. I criticized three strategies the defender of (GG) might employ to avoid this conclusion (Sections 3 ⁴–5). I propose a different strategy: the defender of (GG) should concede that partial pro-attitudes are incorrect.

The success of this strategy depends on the extent to which the defender of (GG) can make the conclusion that partial pro-attitudes are incorrect seem unproblematic. One way to do this is by articulating an analogy for the incorrectness of partial pro-attitudes. If there is some unproblematic phenomenon, to which the incorrectness of partial pro-attitudes is analogous, then our sense that this conclusion is problematic will be mitigated.

Graham Oddie (Reference Oddie2005), in articulating a version of the guise of the good thesis, suggests that partial pro-attitudes are analogous to ‘the kind of systematic ‘illusion’ that is generated by the fact that our visual perceptions are perspectival, ’ as when ‘the sun appears smaller than the moon, even though it is much larger.’ (213; see also Tenenbaum Reference Tenenbaum2007, 39, 56) The illusion, in that case, is constituted by an inaccurate sense perceptual experience – a type of incorrect representation (cf. Section 1). Oddie’s suggestion is that partiality constitutes a comparable species of illusion. My preferring my winning the prize to your winning the prize incorrectly represents my winning the prize as better than your winning the prize, just as my visual experience of the moon and the sun incorrectly represents the moon as larger than the sun. Sergio Tenenbaum (Reference Tenenbaum and Tenenbaum2010), another defender of the guise of the good thesis, suggests that (at least some) partial pro-attitudes constitute a species of ‘malignant recalcitrant illusion, ’ analogous to the case of a superstitious person who consciously rejects the superstition that you can ‘jinx’ your favorite sports team by speaking confidently about their chances of winning, and yet finds it hard to ‘shake off its influence in belief formation, ’ such that they are reluctant to do things that might ‘jinx’ their team. (218) Just as the superstitious person is affected by an incorrect belief in ‘jinxes, ’ the partial person is affected by an incorrect preference that things go well for their friends, as opposed to for others.Footnote ³³

Both of these suggestions are on the right track. However, they fail to capture an important aspect of partiality. Partiality is pervasive, in a way that the illusions described so far are not: partiality is a systematic and ubiquitous feature of our pro-attitudes. Self-love and friendship – again in the broad sense of ‘friendship’ described above (Section 2) – have a constant and wide-ranging influence on what we desire and prefer, such that it is hard to think of a desire or preference that does not in some way manifest our allegiances to ourselves and to our friends. Even our more altruistic pro-attitudes are sometimes affected: your preference to volunteer on Tuesday afternoon as opposed to Friday morning, for example, might be the result of your acquaintance with several of the other regular Tuesday volunteers along with your interest in sleeping in on Friday. However, the illusions described by Oddie and Tenenbaum are not pervasive. The perspectivity of visual experience does not in general involve incorrect representation (Section 4). It is only in exceptional cases – in particular, those in which we cannot accurately perceive relative distance – that perspective gives rise to illusion. This is why, I think, ‘illusion’ must appear in scare quotes when Oddie describes perspectival ‘illusions’ as ‘systematic.’ Likewise, the illusion involved in cases of superstition is an exceptional case of incorrect representation, a single stubbornly irrational belief among a web of beliefs most of which are true.

To capture the pervasiveness of partiality, our analogy must be to a case of pervasive illusion. Consider the view that our visual experiences of color comprise a pervasive illusion, in virtue of their representing things as possessing intrinsic, qualitative properties that they do not possess. When it comes to our visual experiences of color, on the aforementioned view, incorrect representation is not the exception, but the rule. The same, so the defender of (GG) might argue, when it comes to partial pro-attitudes.Footnote ³⁴

However, there is another aspect of partiality that these analogies fail to capture. Partiality is part of our human nature, in a way that the illusions described so far are not. We can easily imagine human beings without visual experiences of color – indeed, many human beings do not have visual experiences of color – just as we can easily imagine human beings not subject to the illusions of visual perspective or the illusions of superstition. But we cannot easily imagine human beings without partial pro attitudes. I do not mean that an impartial human being is impossible, but that an impartial human being would be an exception to the rule: an essentially atypical or abnormal human being.

The analogy we seek, I propose, is partiality in belief. Consider the parent who believes, without sufficient evidence, that their child is the most talented basketball player on the team. This kind of biased thinking is pervasive: our beliefs about ourselves and our friends are typically and in general colored by our emotional attachments to ourselves and our friends. This is a case of pervasive illusion, and one plausibly part of our human nature: we are partial beings, i.e. beings who care more about ourselves and our friends than about others, and our partiality is manifested both by our beliefs and by our pro-attitudes. But our partial beliefs are not, in general, true – these beliefs represent exaggerations about, and overly optimistic interpretations of, ourselves and our friends. Here we have a case of pervasive incorrect representation that is more deeply analogous to the case of partial pro-attitudes.

Above, I distinguished the concept of representational correctness from various prescriptive concepts (Section 1). The case of partiality in belief helps us see why we should insist on this. You might argue that the beliefs of a virtuous person will manifest partiality towards themselves and their friends.Footnote ³⁵ That does not entail that said beliefs are correct. That it is virtuous for you to believe that your child is the best basketball player on the team does not entail that it is true that your child is the best basketball player on the team. Your belief is virtuous even though false. The defender of (GG) thesis should insist on the same point when it comes to partial pro-attitudes. Suppose that the pro-attitudes of a virtuous person will manifest partiality towards themselves and their friends. That does not entail that said pro-attitudes are correct. That it is virtuous for me to prefer my winning the prize to your winning the prize (Section 2) does not entail that my winning the prize is better than your winning the prize.

The conclusion that partial pro-attitudes are incorrect is therefore consistent with ethical defenses of partiality. Consider, for example, the idea that partiality tends to have good consequences and is thus recommended from a consequentialist point of view.Footnote ³⁶ Just as the fact that a belief is incorrect does not entail that it would not have good consequences to adopt it, the fact that a pro-attitude is incorrect does not entail that it would not have good consequences to adopt it. Compare the argument that certain instances of partiality in belief constitute ‘positive illusions, ’ on the grounds that this kind of biased thinking is beneficial and a component of mental health.Footnote ³⁷ The incorrectness of biased beliefs is consistent with this conclusion.

None of this means that the incorrectness of partial pro-attitudes doesn’t matter. Compare, again, partiality in belief. It is surely relevant to the question of whether it is virtuous for you to believe that your child is the best basketball player on the team that such a belief would be false. False belief, in this case, as in many others, has costs, e.g. if your child would benefit from a more accurate critique of their abilities (which the doting parent is unable to provide) or from a training regimen tailored to their deficiencies (which the doting parent does not think necessary). (Perhaps false belief in this case is also intrinsically bad; that would be an additional cost to take into account.) All this matters when it comes to the ethical question of whether it is virtuous for you to believe that your child is the best basketball player on the team. But that ethical question is not settled by the fact that such a belief would be incorrect. The incorrectness of partial pro-attitudes must likewise be taken into account when we address ethical questions about partiality. But those ethical questions will not be settled by the fact that partial pro-attitudes are incorrect.

7. Conclusion

I have articulated the guise of the good thesis (Section 1) and argued that its defenders should maintain that partial pro-attitudes are incorrect (Section 2). I criticized three strategies for avoiding this conclusion (Sections 3 ⁴–5), and sympathetically considered the idea that partial pro-attitudes comprise a species of systematic illusion (Section 6).

In another well-known passage, Hume describes ‘taste’ as having ‘a productive faculty, ’ which by ‘gilding and staining all natural objects with the colours, borrowed from internal sentiment, raises in a manner a new creation.’Footnote ³⁸ If partial pro-attitudes are incorrect, then something like this can be said of partiality: our partial pro-attitudes ‘gild and stain’ ourselves and our friends with the appearance of being more valuable, more important, more significant, than we really are.Footnote ³⁹