Desire-fulfillment welfare theories (DF theories) divide into the actual and hypothetical (or idealized) types. On actual DF theories the agent's well-being is determined by her actual desires, while on hypothetical DF theories the agent's well-being is determined by the desires that she would have if she were fully and vividly informed with respect to non-evaluative information. DF theories are opposed to objectivist welfare theories, which claim that one's own desires (and, more generally, one's own pro-attitudes) do not play any direct role in constituting the aspects of one's own welfare. Objectivist welfare theories often come in the form of objective list theories, which assert (1) that there is an authoritative list of goods (knowledge, accomplishment, health, friendship, aesthetic experience, and pleasure are all apt be on this list) and (2) that something is an aspect of an agent's welfare if and only if, and because, it instantiates one of these goods for her. Although DF theories and objectivist theories are the two most commonly defended types of welfare theories, there are other noteworthy types of welfare theories – namely, welfare hedonism, L. W. Sumner's authentic happiness theory, and hybrid theories that claim that one's welfare is a function both of some non-prudential objective sort of value and of one's own pro-attitudes.Footnote 1 In this article I will assume that we all have on hand an imprecise everyday concept of well-being (and, by extension, an intuitive everyday sense of when it is that people are made better or worse off). But I will not endorse or argue for any particular theory of well-being. My aim is simply to argue against the DF theory view of well-being. My aim, in other words, is simply to show that the DF theory view does not adequately capture or helpfully clarify (or helpfully revise) our imprecise everyday concept of well-being.
My central thesis is that we should reject DF theories because they cannot adequately handle Dead Sea apple cases. Dead Sea apples are apples that look attractive while hanging on the tree, but which dissolve into smoke or ashes once plucked. Accordingly, Dead Sea apple cases are cases in which someone desires something and then gets it, only to find himself disappointed. For example, Dennis might desire a certain job and then get it, only to find that he hates it. And, naturally, we can multiply examples here. Just think of any case in which someone wants something (to take a certain trip, or to join a certain club, or to be in a certain romantic relationship, or to take a certain class, etc.) and then gets it, only to find herself disappointed.Footnote 2
Sidgwick seems to have introduced the phrase ‘Dead Sea apples’ into philosophy. In discussing why he thinks that DF theorists should reject the actual DF theory, he says:
it would still seem that what is desired at any time is, as such, merely apparent Good which may not be found good when fruition comes, or at any rate not so good as it appeared. It may turn out a ‘Dead Sea apple,’ mere dust and ashes in the eating . . .Footnote 3
Sidgwick thought that, if, for each desire that one has, one could be given ‘a perfect forecast, emotional as well as intellectual, of the state of attainment or fruition’, this would help to ensure that one would rid oneself of one's desires for Dead Sea apples.Footnote 4 Thus Sidgwick thought that, if we are going to identify an agent's good with what he or she desires, then we should invoke not the agent's actual desires, but the desires that the agent would have after having been much better informed than he or she now is.Footnote 5
Three sections follow. The first will briefly outline the view of desire that DF theorists typically work with; the second will consider Dead Sea apple cases in relation to actual DF theories; and the third will consider Dead Sea apple cases in relation to hypothetical DF theories.
THE TYPICAL DF THEORY VIEW OF DESIRE
I always use the term ‘desire’ to refer to intrinsic desires (unless I say otherwise). This seems appropriate, since DF theories entail that purely instrumental desires don't directly bear on welfare. As for what desires are, DF theorists typically hold that to desire X (where X is a state of affairs) is to have one's motivational force flowing to X in such a way that one has a tendency to bring X about. With regard to purely instrumental motivations, DF theorists will say that, if one is just instrumentally motivated to bring X about, then one's motivational force isn't flowing to, or directed at, X – rather, one's motivational force is flowing to, or directed at, something that X is thought to be an effective means to. The typical DF theory view of desires as being tendencies to bring about their objects seems to accord with the standard view of the nature of desire. According to Timothy Schroeder, the standard theory of desire says that to desire that P (where P is a proposition) is ‘to be disposed to bring it about that P’.Footnote 6 Also, the standard theory (as described by Schroeder) states:
Desires can play all sorts of causal roles within our heads, but there is just one role definitive of desiring, and that is of engaging the mental machinery in such a way as to tend to bring it about that P, for the P that is the content of the desire.Footnote 7
Some philosophers hold that one can't desire X unless one believes that X is good in some way. But DF theorists typically reject this thesis. The philosophers who advance this thesis ordinarily mean to be claiming that one cannot desire X unless one first cognitively sees that X is good in some way – that is to say, these philosophers ordinarily mean to be claiming that the agent's belief that X is good in some way is what gives rise to her desire for X. But this way of thinking about how we form desires – this desired because first cognitively seen to be valuable way of thinking – seems antithetical to the DF theory way of thinking about how we form desires. After all, on DF theories our desires make things be prudentially good for us in the first place. Thus, according to DF theories, the correct order is (prudentially) valuable because desired.Footnote 8 To be clear, there is room for DF theorists to accept the thesis that one cannot desire X unless one believes that X is good in some way. The idea here is that a DF theorist could say: ‘Things are prudentially good for you because, and only because, you desire them. But still, your desire for an object always arises at least in part because you see some sort of non-prudential desire-independent goodness in it.’ At least typically, though, DF theorists allow that one can desire X without believing that X is (in any way) good. Or, spelled out more fully, DF theorists don't typically claim that our desires have the function of tracking what is independently valuable – rather, what DF theorists typically claim is that our desires are such that they spontaneously fix on certain value-free objects and thereby make them be prudentially valuable.
DEAD SEA APPLE CASES IN RELATION TO ACTUAL DF THEORIES
I am aware of three responses that actual DF theorists might advance when confronted with Dead Sea apple cases: the fine-grained response, the net loss response, and the concurrence response (which derives from Chris Heathwood).Footnote 9 Let's take these three responses one by one.
The fine-grained response
The objector to the actual DF theory might say this. ‘Dennis wants to work at a certain law firm and then gets the job, only to find that he hates it. Here his desire is fulfilled, but his well-being is not advanced. It follows that the actual DF theory fails.’
In response to this Dead Sea apple objection, DF theorists might attempt to fine-grain Dennis's desires. DF theorists might say: ‘The case has been misdescribed. In truth the case runs as follows. There are various pros and cons to Dennis's job. The pros are the interesting cases that Dennis sometimes works on, those few pleasant co-workers of his, etc. And the cons are the long hours, the boring cases that he sometimes works on, his super-cranky boss, etc. Dennis likes the pros of his job and dislikes the cons of his job. Also, we know that Dennis on the whole hates his job, and presumably this is so because the various aspects of his job that he dislikes somehow greatly outweigh the various aspects of his job that he likes. Does it make sense to think that (prior to getting the job) Dennis ever had a desire for any of the various aspects of his job that he now dislikes (working long hours, having a super-cranky boss, etc.)? No, it doesn't. At least on that score, then, there is no problem with Dennis's desires. Further, it makes sense to think that Dennis never desired the state of affairs his working at this law firm (full stop). Rather, what he desired (prior to getting the job) were just certain aspects of the job (to work on interesting cases, to work with pleasant people, etc.). Dennis was motivated to take the job, then, not because he desired to take the job (full stop), but rather because he desired certain aspects of the job and because he knew that he couldn't get these aspects without taking the job. The bottom line here is that this case doesn't involve any desires that, when fulfilled, fail to advance Dennis's welfare.’
I will now make two points about the fine-grained response outlined above. First, one problem, or possible problem, with this response is that it undermines the typical DF theory view of desire (which entails our desires are such that they fix on value-free objects). In effect, the fine-grained response says: ‘People never desire things like having such-and-such a job or being in such-and-such a romantic relationship. All that people ever desire are certain aspects – indeed, certain positive aspects – of having such-and-such a job and of being in such-and-such a romantic relationship.’ According to the fine-grained response, then, our desires always zero in on the positive, and only the positive, aspects of the experiences that are open to us. Yet, if this is the right view of desire, then our desires no longer seem to be such that they spontaneously fix on value-free objects – rather, our desires seem to be such that they reliably track independent values. Second, the fine-grained response is implausible in itself. It is meant to cover all putative Dead Sea apple cases. But this goes too far. Yes, there are plenty of cases where people pursue certain jobs, certain relationships, etc., just because they desire certain aspects of these jobs, relationships, etc. But it goes too far to say that people never desire to have certain jobs (full stop), that people never desire to be in certain relationships (full stop), etc. And in fact the norm seems to be that people desire to have the job (full stop), to be in the relationship (full stop), etc.
Given, though, that it is true that we typically desire to have the job (full stop), to be in the relationship (full stop), etc. – as opposed, that is, to just desiring certain aspects of the job, the relationship, etc. – there must be a reason for this. But what is this reason? It seems that in some cases we form coarse-grained desires (such as the desire to have the job, full-stop) simply because we fail to consider all of the relevant details of the situation. If, in these cases, we were to consider all of the relevant details of the situation, then we would form nothing but fine-grained desires for certain positive aspects of the situation. Yet there seems to be more to the matter than just this. After all, it seems that it often happens that we form coarse-grained desires even in spite of our being aware of all of the relevant details of the situation. To make sense of this matter, we can consider something that the psychologist Daniel Gilbert has said with reference to memory:
How do we cram the vast universe of our experience into the relatively small storage compartment between our ears? . . .We cheat . . . [T]he elaborate tapestry of our experience is not stored in memory – at least not in its entirety. Rather, it is compressed for storage by first being reduced to a few critical threads, such as a summary phrase (‘Dinner was disappointing’) or a small set of key features (tough steak, corked wine, snotty waiter). Later, when we want to remember our experience, our brains quickly reweave the tapestry by fabricating – not by actually retrieving – the bulk of the information that we experience as a memory. This fabrication happens so quickly and so effortlessly that we have the illusion . . . that the entire thing was in our heads the entire time.Footnote 10
Here let's focus on Gilbert's point about how we take the rich, detailed tapestry of our experience and then distill out of it a few critical threads, so that these few critical threads can then be stored in our memory banks. A similar point can be made with respect to desire formation. We seem to have a pragmatic need to keep ourselves from being overwhelmed by the complexity or richness of the world. Yes, our brains can store a remarkable amount of information; and, yes, we have an amazing ability to keep track of details. But still, if we were unable to reduce the experiences open to us to certain summary judgments (say, the judgment that this job is a good job), then we would be overwhelmed – that is to say, we would be flooded by the details of these experiences. My point here, then, is this. Even in cases where we are aware of all of the relevant details of the experiences open to us, we have a tendency to form coarse-grained evaluative beliefs concerning these experiences (say, the belief that this job is a good job) and to form coarse-grained desires that are based on these coarse-grained evaluative beliefs (say, the desire to get this job). These coarse-grained beliefs (and the coarse-grained desires that are based on them) have the function of generalizing and systematizing information in a way that is pragmatically crucial for us.
The net loss response
When confronted with Dead Sea apple objections, DF theorists might say this. ‘Go back to Dennis. By working at the law firm Dennis fulfills some desires – for instance, the desire to meet some nice co-workers and also the desire to work at the firm (full stop). Inasmuch as these desires are fulfilled, Dennis's welfare is advanced. However, by working at the law firm, many of Dennis's other desires are frustrated – for instance, the desire to have good hours, the desire to work on certain types of cases, and the desire to have a nice boss. Inasmuch as these desires are frustrated, Dennis's welfare moves backward. Once all of the relevant desire-satisfactions and desire-frustrations are taken into account, it will be clear to us that, as far as this case goes, Dennis's total desire-frustration outweighs Dennis's total desire-satisfaction. The key point here, then, is this: Inasmuch as Dennis's desire to work at the firm (full stop) is fulfilled, his welfare is advanced; and the net loss in welfare that Dennis suffers as a result of working at the firm can be explained by DF theories without any problem, for this net loss can be explained in terms of total desire-frustration outweighing total desire-satisfaction.’
This response works only if we grant that the desires being frustrated here are, on balance, stronger than the desires being fulfilled here. But why, exactly, should we grant this? James Griffin says: ‘It is depressingly common that when even some of our strongest and most central desires are fulfilled, we are no better, even worse, off.’Footnote 11 What if Dennis's desire to work at the firm (prior to his getting the job) was an extremely strong or deeply central desire? Would it still be clear, in that case, that DF theorists are justified in claiming that the net loss in welfare that Dennis suffers as a result of working at the firm can be explained in terms of total desire-frustration outweighing total desire-satisfaction?
Suppose, though, that we can leave this issue of desire-strength assumptions to the side. Even then, there is still a major problem with the net loss response. As applied to the case with Dennis, the net loss response says that Dennis's working at the firm is something that in itself adds to Dennis's welfare. Indeed, the claim here is that, inasmuch as Dennis's desire to work at the firm (full stop) is fulfilled, his welfare is advanced. Yet, unless one has a prior commitment to the DF theory, there is no good reason to accept this claim. After all, Dennis is miserable at the firm. Dennis and his job are not at all suited or fit for each other. Thus, unless one is already a DF theorist, one will inevitably conclude that working at the firm is not in itself adding to Dennis's welfare. DF theorists need to explain why this natural conclusion is mistaken. But, on reflection, there seems to be nothing constructive that DF theorists can say here. Naturally, DF theorists might say that it is just a fact that all instances of desire-fulfillment add to welfare. But that won't be convincing to anybody who isn't already a DF theorist.
The concurrence response
In dealing with Dead Sea apple cases, actual DF theorists might advance the concurrence response, which derives from Heathwood. Heathwood says:
Importantly, I mean the theory above [the actual DF theory sketched above] to require concurrence: in order for a state of affairs to count as a genuine instance of desire satisfaction, the state of affairs desired must obtain at the same time that it is desired to obtain. If I desire fame today but get it tomorrow, when I no longer want it, my desire for fame was not satisfied. A desire of mine is satisfied only if I get the thing while I still desire it, and continue to have the desire while I'm still getting it.Footnote 12
With respect to the case involving Dennis, Heathwood or any other defender of concurrence may say: ‘DF theorists need not worry about this case, since it does not satisfy the concurrence requirement. After all, the state of affairs that Dennis works at the law firm does not obtain at the same time that Dennis has the desire for this state of affairs. Yes, before Dennis works at the law firm, Dennis has the desire for this state. However, by the time Dennis works at the law firm, he no longer has the desire for this state (since, after all, he hates his job).’ Heathwood is very forthcoming about why actual DF theorists should embrace concurrence. He says:
Oscar Wilde quipped that the only thing worse than not getting what one wants is getting what one wants. Sidgwick lamented a related point, noting that the objects of our strongest desires often come to us as ‘Dead Sea apples’, no longer wanted once they are gotten, ‘mere dust and ashes in the eating’. Sidgwick thought that moving to an ideal desire theory, one requiring full information about the objects of desire, could solve the problem, since we could learn in advance whether the objects of our desires would come to us flat . . . But actualist desire-satisfaction theories can accommodate the Dead-Sea-apple phenomenon. They can do so by building concurrence into the theory, as I have done above. Genuine desire satisfaction is had only when the desire remains once its object is gotten. The concurrence requirement ensures that the getting of Dead Sea apples doesn't improve welfare, since the very reason the thing is a Dead Sea apple is that the desire for it has vanished.Footnote 13
I will now discuss two problems with the concurrence response to Dead Sea apple cases: the overlap problem and the merely near-possible problem.
The overlap problem
With respect to some Dead Sea apple cases, it seems that there will be a stretch of time, even if only a short one, during which the agent in question is still desiring X, is in fact getting X, and is not benefiting from X. For instance, when Dennis first starts his job at the firm, it may happen that he still has the desire to work at the firm, that his desire is being fulfilled (since he is in fact working at the firm), and that he is not benefiting from working at the firm.
This point can be brought out more clearly if we consider that Dennis, looking back on his time spent working at the firm, might say: ‘I worked at the firm for a month. The whole experience was terrible. I was miserable there. Before I got there, I wanted to work there. Also, I would say that I still wanted to work there during my first week on the job. After that first week, though, I lost my desire to work there. I stayed the rest of the month just so as to make sure that I had an income until I found a new job. Why did I still want to work at the firm during my first week there? That is a good question, for I was having a terrible time at the firm, even during that first week. I initially formed the desire to work at the firm because I believed that this job was an attractive job. And I continued to hold this defective belief even during my first week of working at the firm. Indeed, it wasn't until after this first week that it finally hit me that this job was in fact an unattractive job. Things just didn't crystallize in my mind until after that first week on the job. Thus it wasn't until after this first week that I lost my desire to work at the firm. Anyway, it is clear to me now, looking back, that I never benefited from working at the firm, not even during that first week when I still wanted to work there (or, as you philosophers might put it, when I still had some of my motivational force flowing to the state that I work at the firm).’
While it is obviously wrong to think that all Dead Sea apple cases involve an overlap problem, it does seem sensible to think that a fair number of Dead Sea apple cases involve an overlap problem. And here, of course, by ‘overlap problem’ I mean ‘a stretch of time during which the agent continues to want X even as he or she is getting X and is not benefiting from it’.
The merely near-possible problem
When Sidgwick, Griffin, and others advance the Dead Sea apple line of objection, they are thinking that a person's actual desires can and do aim at things that are not intrinsically good for her, and that we can see that this is so when we reflect on the fact that a person sometimes gets what she actually wants, but finds that this thing is disappointing to her. The defender of concurrence answers: ‘But, since she doesn't want the thing once she gets it, the DF theory is in the clear here – or, at any rate, the DF theory is in the clear if we build concurrence into the DF theory.’ Yet it seems that the claim that Sidgwick, Griffin, and others are making still stands: Our desires can and do aim at things that aren't intrinsically good for us, and Dead Sea apple cases nicely bring this out. With respect to Dead Sea apple cases, we should say: Even if someone's desire for X (which is not intrinsically good for her) disappears the moment that she actually gets X, this doesn't change the fact that her desires were going wrong (or were aiming at something – namely, X – that was not intrinsically good for her) just before she got X. Here the defender of concurrence is apt to protest: ‘But, when X obtains, she doesn't desire X. So there is no problem with her desires then (that is, when X obtains). If there is any problem with her desires here, the problem lies with her desire for X prior to X's obtaining. But can't we just bracket states of affairs of the non-obtaining sort?’
The above remarks suggest that we need to think about non-obtaining states of affairs in relation to DF theories. Here is one way – and, to be clear, this is an anti-concurrence way – of thinking about this matter. DF theories reduce well-being to desire, and so on DF theories states of affairs become aspects of one's welfare the moment that one desires them (it doesn't matter whether the desired states of affairs obtain or not). Thus, if Nikhil now desires the state of affairs Nikhil's drinking/metabolizing milk, then this state of affairs is now prudentially good for Nikhil (and it is so whether it obtains or not). Continuing with this line of thought, we might draw a distinction between near-possible and far-possible aspects of welfare. Take A to be a human and X to be a state of affairs. ‘X is a near-possible aspect of A's welfare’ should be taken to mean ‘X is now prudentially good for A’. The near-possible aspects of one's welfare fall into the merely near-possible and the actual types. ‘X is a merely near-possible aspect of A's welfare’ should be taken to mean ‘X is now prudentially good for A, but X doesn't now obtain’. ‘X is an actual aspect of A's welfare’ should be taken to mean ‘X is now prudentially good for A, and X now obtains’. ‘X is a far-possible aspect of A's welfare’ should be taken to mean ‘X isn't now prudentially good for A, but X might later come to be prudentially good for A’. With respect to DF theories, these distinctions should be understood as follows. If one desires X, then X is a near-possible aspect of one's welfare (that is to say, X is prudentially good for one). If one desires X but X doesn't now obtain, then X is a merely near-possible aspect of one's welfare. If one desires X and X now obtains, then X is an actual aspect of one's welfare. If one doesn't desire X but might desire X at some point in the future, then X is a far-possible aspect of one's welfare.
Let's suppose that we accept these distinctions, if only provisionally. And let's now return to the case with Dennis and try to pinpoint why it spells trouble for DF theories. Here, moreover, let's suppose that there is no overlap problem (that is to say, let's suppose that Dennis loses his desire to work at the firm the moment that he actually starts working there). My aim here is to show that, even when there is no overlap problem for DF theorists to worry about, the concurrence response still seems like an inadequate response to Dead Sea apple cases. Suppose, then, that at t1 Dennis forms the desire for the state that he works at the firm. It follows on DF theories that this state is a merely near-possible aspect of Dennis's welfare at t1. But that is hard to swallow. In order for this state to be a merely near-possible aspect of Dennis's welfare at t1, this state must be prudentially valuable for Dennis at t1. But why should we think that this state is prudentially valuable for Dennis at t1? We know that, as things turn out, Dennis works at the firm from t2 indefinitely onwards. And we also know that, as things turn out, Dennis is miserable while he works at the firm. In other words, we know that, when this state does obtain at t2, it is not prudentially valuable for Dennis. But, if we know that this state is not prudentially valuable for Dennis when it obtains at t2, then why should we think that this state is prudentially valuable for Dennis at t1 (that is, just prior to its obtaining)? There really is no good reason to think that this state is prudentially valuable for Dennis at t1. If I am right about this, then DF theories have this case wrong – that is to say, DF theories here wrongly entail that this state is prudentially valuable for Dennis at t1 (or is a merely near-possible aspect of Dennis's welfare at t1).
Here the defender of concurrence will protest: ‘But, if we build concurrence into DF theories, then DF theories don't entail that this state is prudentially valuable for Dennis at t1. After all, this state doesn't obtain at t1, and the concurrence requirement says that the only states that can be prudentially valuable for an agent are those states that the agent desires and that obtain at the same time that the agent desires them. Speaking generally, if DF theorists accept concurrence, then DF theorists can simply reject the claim that there are such things as merely near-possible aspects of welfare. And, once DF theorists do that, they can reject the claim that DF theories here wrongly entail that the state Dennis's working at the firm is prudentially valuable for Dennis at t1.’ This response urges DF theorists to embrace concurrence, and so to reject the claim that there are such things as merely near-possible aspects of welfare. But in fact it seems that DF theorists need to countenance merely near-possible aspects of welfare.
To see why I say this, suppose that Mary is seventeen years old, that she now desires to be an architect by the age of twenty-six, and that she does not now desire to be a veterinarian by the age of twenty-six. Given that we are already accepting a DF theory framework, it seems quite clear that Mary's prudential relationship with the state her being an architect at age twenty-six importantly differs from Mary's prudential relationship with the state her being a veterinarian at age twenty-six. How are DF theorists going to account for this difference? One straightforward move here is for DF theorists to say that, right now, it so happens that the state Mary's being an architect at age twenty-six is a merely near-possible aspect of Mary's welfare, while the state Mary's being a veterinarian at age twenty-six is a far-possible aspect of Mary's welfare. DF theorists seem to be faced with a choice. They can countenance merely near-possible aspects of welfare, in which case they can explain what is going on in cases such as the case with Mary – or, alternatively, they can refuse to countenance merely near-possible aspects of welfare, in which case they are going to have serious difficulties explaining what is going on in cases such as the case with Mary. Or, put more generally, my point is this. Given that we are already assuming a DF theory structure, it seems clear that desired states of affairs, even if they don't now obtain, should be treated as having a different kind of importance vis-à-vis well-being than non-desired states of affairs. The problem with the concurrence requirement is that it puts desired states of affairs that don't now obtain on the same level of importance as non-desired states of affairs.
Here defenders of concurrence might protest: ‘Look, we can embrace concurrence and, in addition, claim that those states that A desires but that don't now obtain are importantly different from those states that A doesn't now desire. We can say that the former states are close to being prudentially good for A, while the latter states are not.’ This response is inadequate. Though it will get the right result in many cases, it will also fail in many cases. Here consider Dennis again. At t1 Dennis is not yet working at the firm, but he does desire to work there. Then at t2 Dennis is working there and is miserable. Defenders of concurrence may claim that at t1 the state Dennis's working at the firm is among those states that are close to being prudentially good for Dennis. But this seems false. After all, at t1 the state Dennis's working at the firm doesn't seem close to being prudentially good for Dennis. I say this because, when this state does obtain soon after (at t2), this state is clearly not prudentially good for Dennis (since Dennis is miserable at the firm). We should not believe, then, that this state really is close to being prudentially good for Dennis at t1.
In sum, the concurrence response to Dead Sea apple cases seems inadequate for two reasons. First, it is plagued by a certain overlap problem; and, second, it seems incapable of adequately handling desired states of affairs that do not now obtain.
To sum up for this section, all three of the above responses (the fine-grained response, the net loss response, and the concurrence response) seem to be, all things considered, inadequate. And, since these seem to be the only available (or at least the best available) actual DF theory responses to Dead Sea apple objections, it seems that we should reject actual DF theories.
DEAD SEA APPLE CASES IN RELATION TO HYPOTHETICAL DF THEORIES
The idea behind hypothetical DF theories is this. Take an agent. Give her all of the non-evaluative facts there are. Make sure, insofar as possible, that she vividly internalizes these facts; and also make sure, insofar as possible, that she is free of cognitive error and that she is being instrumentally rational. When this process is over, she will have various beliefs and desires. The desires that she has at the end of this process determine her well-being.Footnote 14
To be clear, the point of fully and vividly informing the agent is, in the first instance, to get her to change her beliefs. The point is to get her to criticize her beliefs in light of the facts. The change in her desires is supposed to be effected through the change in her beliefs.Footnote 15
With respect to Dead Sea apple cases, then, hypothetical DF theorists need to ensure that fully and vividly informing agents would cause them to change their beliefs and then in turn to change their desires such that they no longer desire Dead Sea apples. With respect to Dennis, this means that hypothetical DF theorists need to ensure that fully and vividly informing Dennis would cause him to believe that the job at the firm is not an attractive job. If fully and vividly informing Dennis were to cause Dennis to believe that the job at the firm is not an attractive job, then it stands to reason that Dennis would lose his desire to work at the firm. Accordingly, the key question to ask here is whether fully and vividly informing Dennis (prior to his taking the job at the firm) would cause him to believe that the job at the firm is not an attractive job.
As Dennis is being fully and vividly informed, he will be told various value-free facts concerning his potential boss, concerning the number of hours that people in his potential position have worked in the past, concerning his potential co-workers, etc. Once Dennis has been made aware of all of the relevant value-free facts, and once he has vividly internalized all of these facts, it may happen that he forms the belief that the job at the firm is not an attractive job; and in turn, then, it may happen that he loses his desire to work at the firm. This is, of course, the result that hypothetical DF theorists are looking for. Yet there seems to be no guarantee that hypothetical DF theorists will get this result. One point to stress here is that the process of fully and vividly informing someone is intended to be purely causal: There are no normative/evaluative constraints loaded into the process. The agent is supposed to be given a bunch of value-free facts; the agent is supposed to internalize these facts vividly; and then, once this is over, the agent will have various beliefs and desires, many of which may be defective from the standpoint of common sense. It is the desires that the agent winds up with – however defective they may be from the standpoint of common sense – that are supposed to determine the agent's welfare.
With respect to Dennis, then, there seems to be no way for hypothetical DF theorists to guarantee that it won't happen (1) that at t1 Dennis has on hand all of the relevant facts, (2) that at t1 Dennis believes that the job at the firm is an attractive job and that things will go a certain way if he works there, (3) that at t1 Dennis desires to work at the firm (at least in part because he believes that the job at the firm is an attractive job), and (4) that at t2 Dennis is working at the firm and finding, much to his dismay, that things are not going as he had expected them to go. Or, spelled out in more detail, we might say this. At t1 Dennis has on hand all of the relevant facts. As Dennis is vividly internalizing all of these facts at t1, he naturally puts more weight on some of these facts, and less weight on others. For instance, Dennis puts a good deal of weight on the fact that, in the past, the firm has dealt with many cases of a type that Dennis finds very interesting; and Dennis puts less weight on the fact that the people who have worked in his potential job position have averaged sixty-hour work weeks for the last five years. At t2, then, Dennis is working at the firm and finding, much to his dismay, that the firm is not working on as many of the types of cases he likes to work on, and that he is working seventy-plus hour work weeks. It is not as though Dennis didn't foresee these possibilities; he was aware that things might turn out this way. Still, he had expected things to go differently, and he is disappointed that things aren't turning out as he had expected. In spelling all this out, we might add that Dennis didn't make any cognitive errors when he formed his beliefs about how things would go.
In response to all of this, hypothetical DF theorists may say: ‘The above remarks assume that, as Dennis is being informed, he will only be given facts about how things have gone in the past and facts about how things stand in the present. However, as Dennis is being informed, he will also be given facts about how things would go for him if such-and-such were to happen. For example, then, as Dennis is being informed, he will be given facts about how things would go for him, phenomenologically and otherwise, if he were to take the job at the firm. If Dennis is given facts such as these, and if it is thereby made clear to Dennis that he would be miserable if he were to work at the firm, then this would guarantee that fully and vividly informing him would cause him to believe that working at the firm is not an attractive job (and this in turn would cause him to lose his desire to work at the firm).’ Here, though, we should stop and ask whether there really are any facts about how things would go for someone if such-and-such were to happen.
David Sobel thinks that it is quite clear that there are facts of this sort. He says that, in providing an agent in the hypothetical setting with more information than she actually has, the goal is ‘to provide her with a more accurate understanding of what the option she is considering would really be like.’Footnote 16 And, while elaborating on this matter, Sobel says:
This idealization is an attempt to get already existing facts into better focus . . . [T]he facts it attempts to get into better focus are the nonnormative facts about what it would be like to have various ways one's life might go be actual. There is a fact of the matter about this. The idealization is an attempt to get us to see such facts clearly.Footnote 17
Here is how Sobel argues. He starts with a simple case, an ice cream case, where (by his lights) it is very clear that there are facts about what it would be like for oneself if one were to eat different flavors of ice cream. Then Sobel generalizes (or analogizes) from this case. That is, Sobel seems to think that, in light of this ice cream case, we are justified in thinking that there is a general truth to the effect that there are facts about what it would be like to have various ways one's life might go be actual. Two questions arise here. (1) Does the ice cream case really work? (Is it clear that there are facts about what it would be like for someone if she were to eat different flavors of ice cream?) (2) Given that the ice cream case works, are we justified in generalizing from this case?
With respect to the first question, there are good reasons for thinking that the ice cream case works. Here consider what Sobel says:
Think about a careful test of which flavor of ice cream one likes best. One might, and experts no doubt do, devise a raft of idealizing procedures to help in this task. One might hold that a person's palate should be cleansed between each flavor being assessed. Or one might hold that the person should be comparably hungry when tasting each flavor or that the order of presentation of flavors be changed to make sure that things such as the lingering effect of toothpaste on one's palate is not affecting the flavor of the first taste and so on. All of this is designed to give the agents more accurate information about what it would be like for them to eat this or that ice cream.Footnote 18
We do tend to think that there are facts of the matter as to what it would be like for us if we were to eat various ice cream flavors. Moreover, we do tend to think that, in order to get clear on these facts, it is best to do away with whatever might interfere with our ability to know these facts. If Bobby were to eat a bunch of super-hot jalapenos and then, right after that, to try some sort of ice cream that he has never tried, and if he were, after that, to say to us that it's not clear to himself whether he likes or doesn't like this sort of ice cream, we would not lose our belief that there is a fact of the matter as to whether he likes or doesn't like (or is indifferent toward) this sort of ice cream. What we might say is this: ‘Bobby, let's wait a few days – say, till your taste buds have grown back – and then we'll have you eat this sort of ice cream again. Then, as long as the conditions are right (for instance, as long as you haven't been eating jalapenos again), it will become clear to you, and in turn to us, whether you like this sort of ice cream or not.’
But can we generalize from this ice cream case? To focus the issue here, we can consider Sidgwick's discussion of an agent's being given a perfect forecast of the state of attainment for the various objects that she desires. With this discussion of Sidgwick's in mind, Sobel says:
Only desires formed in light of an accurate understanding, phenomenological and otherwise, of what an option would be like are responsive to the true nature of the object under consideration . . . The point of the idealization is to give an accurate understanding of what the option that one is assessing is like. The best rationale for the idealization is that it promises this perfect forecast. To the extent that an idealization cannot promise this, it loses this strong and intuitive rationale.Footnote 19
As I have indicated, we should grant that it is possible for someone to be given a perfect forecast of what it would be like for her to eat this or that ice cream flavor. The key question to ask here, though, is this: Is it really possible, for all or almost all of the desires that someone has, to provide her with a perfect forecast of what it would be like for her to have these desires fulfilled?
I am now going to argue that this is not possible. This is not possible because many of the objects that someone desires are, in terms of their very natures, subject to too much flux or instability to allow for the providing of the sort of perfect forecast that is in question here. Notice that an ice cream flavor is both a relatively simple and a relatively stable sort of thing. Its nature is not particularly complex and is not subject to any significant flux. Moreover, while someone's palate is subject to some flux (for instance, the child that doesn't like vegetables may become the adult who does like vegetables), the fact remains that someone's palate is apt to stay fairly constant, at least for certain stretches of time. Thus there seems to be the space that is needed for a fairly stable relationship to develop between an ice cream flavor and someone's palate. It is within this space that we find settled facts as to what it would be like for someone if she were to eat this or that ice cream flavor. But now turn to the case with Dennis. Can we isolate any fact about what it would be like for Dennis if he were to work at the firm? In order to be able to isolate any such fact, we must be able to hold Dennis's personality fairly constant, and we must also be able to hold the nature of the job at the firm fairly constant (as otherwise there won't be the space that is needed for a fairly stable relationship to develop between Dennis and the job at the firm). Even if we grant that holding Dennis's personality fairly constant isn't a problem, it is hard to see how we can hold the nature of the job at the firm fairly constant. I say this because there are so many variables in play here: the number of hours that Dennis works, the moods of Dennis's boss and co-workers, the moods and temperaments of the various clients for whom Dennis works, the types of cases that Dennis works on, the pay-cuts and/or pay-raises and/or bonuses that Dennis may receive, the traffic on the way to work each day, the amount of traveling that Dennis is required to do, etc. It seems that none of these variables truly counts as being an interference in the way dulled taste-buds resulting from the eating of super-hot jalapenos truly count as being an interference in the ice cream case. Indeed, all of these variables seem directly to bear on and/or to be intrinsic to the nature of Dennis's job at the firm, and therefore all of these variables seem to be in play when we are asking about how things would go for Dennis if he were to work at the firm. However, all of these variables are subject to a good deal of flux or instability; they certainly aren't stable in the way that the nature of an ice cream flavor is stable.
The upshot of these remarks is this. It seems that, prior to Dennis's working at the firm, there is no set fact as to what it would be like for Dennis if he were to work at the firm. This is so because, depending on various factors (the number of hours that Dennis works, the types of cases he works on, etc.) and their interplay with each other, things could be quite nice for Dennis if he were actually working at the firm – or, alternatively, things could be far from nice for Dennis if he were actually working at the firm. It follows that, prior to Dennis's working at the firm, Dennis – when he is in the hypothetical setting – cannot be told that, if he were to work at the firm, he would be miserable (for this is not a fact). But, if Dennis is not told this when he is in the hypothetical setting, then it is unlikely that he will come to believe that the job at the firm is not an attractive job (and in turn it is unlikely that he will lose his desire to work at the firm). It follows, then, that hypothetical DF theories probably cannot handle this case. (And this is but one of many Dead Sea apple cases that pose serious problems for hypothetical DF theories.)
Here hypothetical DF theorists might say: ‘It is true that, before Dennis works at the firm, there is no set fact as to what things would be like for him if he were to work there. But there are set facts like the following one: If Dennis were to work at the firm, and if (while working there) he would be focusing on cases of such-and-such a sort, working seventy or more hours in a week, dealing with a super-cranky boss, etc., then he would be miserable. When Dennis is being fully informed, he will be told set facts of this sort. And, once he is told all of the set facts of this sort, it stands to reason that he will no longer desire to work at the law firm (full stop) – rather, what he will desire is to work at the law firm, provided only that such-and-such conditions obtain (say, that he is working on cases of a certain type, that his hours aren't insane, that his boss is relatively easy to deal with, etc.). In sum, then, Dennis will refine, or fine-grain, his desires in light of full information; and, once he does this, this case won't be a problem for hypothetical DF theories.’
This response is not persuasive. Earlier I noted that, even in cases where we are aware of all of the relevant details of the experiences open to us, we have a tendency to form coarse-grained evaluative beliefs concerning these experiences (say, the belief that this job is a good job) and in turn to form coarse-grained desires that are based on these coarse-grained evaluative beliefs (say, the desire to get this job). We have this tendency because forming these coarse-grained beliefs and coarse-grained desires helps us in satisfying our pragmatic need not to be overwhelmed by details (or by the complexity of our experience of the world). This pragmatic need is apt to become even more pressing when we are in the hypothetical setting. This is so because, when we are in the hypothetical setting, we will be bombarded with zillions of facts. In order to ensure that we are not overwhelmed by all of this information, we will need to reduce much of this information to certain summary beliefs or judgments (say, the summary belief that this job is a good one); and, once these coarse-grained, summary beliefs are in place, coarse-grained desires that are based on these beliefs (say, the desire to have this job) will follow behind.
An example may help here. Suppose that Lisa is searching for philosophy jobs. She is aware of about fifty jobs for people in her field. She is very thorough in her approach to the job search. For each job, she learns about the pros and cons that attach to it. For instance, for school Q, Lisa knows (1) that the department at school Q is good in her field, which is a plus; Lisa knows (2) that the undergraduates at school Q are typically smart and motivated, which is a plus; Lisa knows (3) that the teaching load at school Q is usually manageable, which is a plus; Lisa knows (4) that school Q is up north, which is a minus, since Lisa doesn't like cold winters; Lisa knows (5) that there is a high cost of living in the city in which school Q is located, which is a minus; and on this goes for more pluses and minuses. And the point here is general: Lisa has an in-depth knowledge not only of the job at school Q, but of all of the jobs in question here. Let's now ask about the desires that Lisa will have here. Will Lisa have nothing but super-fine-grained desires (or desires with various conditions attached)? Regarding school Q, for example, will Lisa have the desire to get the job there, provided only that various conditions are met – say, provided only that the professors that are good in her field don't leave for other schools, that school Q doesn't unexpectedly up the teaching load, that she is able to find an inexpensive apartment, etc.? It is plausible to think that, with respect to some of the jobs in question here, Lisa will form super-fine-grained desires. But it seems implausible to think that Lisa will form nothing but super-fine-grained desires here. There are fifty or so jobs in question; and, for each of them, there are plenty of details to keep track of. Even if Lisa is very thorough and precise in her approach to this job search, she will still need to reduce the complexity of the information that she has on hand here, since otherwise she will be overwhelmed by it. The obvious way to do this – and the way that we do this in real life – is to reduce much of this information to coarse-grained beliefs (say, the belief that the job at school R is good); and, once these coarse-grained beliefs are in place, coarse-grained desires that are based on them will follow (say, the desire to get the job at school R).
Like Lisa, agents in the hypothetical setting will need to reduce the complexity of the information that is coming at them. In fact, since agents in the hypothetical setting will deal with so much more information than Lisa is here dealing with, their need to reduce the complexity of the information coming at them will very likely be even more pressing than Lisa's need to reduce the complexity of the information coming at her here. It stands to reason, then, that agents in the hypothetical setting will form a great many coarse-grained beliefs (say, the belief that this job is a good job) and a great many coarse-grained desires that are based on these coarse-grained beliefs (say, the desire to get this job). Also, just as many of the coarse-grained actual desires that agents have are desires for Dead Sea apples, so too many of the coarse-grained desires that agents will have after having just been fully and vividly informed will be desires for Dead Sea apples. The upshot of this is that hypothetical DF theories can't adequately handle Dead Sea apple cases.
WHERE DOES THIS LEAVE US?
At first glance, it seems that we often desire things that, when gotten, make us worse off (say, to have a certain job, or to be in a certain relationship). DF theorists think that, on deeper reflection, we will see that DF theories can handle Dead Sea apple cases. But, if I am right, this is not true. If we reject DF theories, this leaves objectivist theories, welfare hedonism, Sumner's theory, and hybrid theories as our main remaining welfare theory options. I will leave it to the reader to judge for himself or herself as to which of these remaining options seems best, where by ‘best’ I mean not only ‘best with regard to Dead Sea apple cases’, but also ‘best in general’.Footnote 20
