I value things. Most of us do. But what is it to value something? Am I having an emotional reaction, a belief, a desire or some combination thereof? The desire option is not without prominent adherents. Davidson[i] and Gauthier[ii] both advocated the idea that valuing can be reduced in some way to desiring. Lewis advocated, and even Russell briefly flirted with, the idea that valuing is merely desiring to desire.[iii] Nevertheless, these prominent reductions are vulnerable to fairly obvious counterarguments. Compulsiveness, weakness of will, depression, valuing without desiring, arbitrariness, vicious regressions of higher-order desires and other such possibilities militate against theories attempting to reduce valuing to desiring. Though I am currently ambivalent as to whether or not such a reduction is in the end likely to be correct, I do find that there is a strong intuitive pull to the idea that my valuing something may be nothing more than desiring it in some way. In particular, it seems to me that some version of a higher-order theory of desire could be a serious contender in the debate, provided that it is sophisticated enough to withstand some of the more obvious criticisms.
My goal in this paper is to sketch a plausible higher-order desire theory of valuing. For this task I rely heavily on David Lewis’s seminal dispositional theory of value and corresponding criticisms by Mark Johnston and Michael Smith. In the first section I briefly discuss the problems that any theory of valuing must address, and the initial plausibility of Lewis’s theory in addressing them. I then outline criticisms that are destructive against Lewis’s theory. Section II is for the most part dedicated to the primary goal of sketching a higher-order desire theory that will plausibly match our intuitions about value while appropriately handling the criticisms leveled against Lewis. In section III, I put my theoretical sketch under a bit of self-scrutiny in order to further refine and hopefully insulate it against at least one possible criticism.
I: Valuing
“To be valued by us means to be that which we desire to desire,”[iv] says David Lewis. Ultimately, the definition fails. Lewis passes over a variety of problems far too quickly, which inevitably leads to oversights in his definition of valuing; some remediable, some not. But it’s not bad as a basis for exploration, and it certainly has its virtues. It addresses a couple of requirements of any viable theory of valuing. The first is that valuing seems to be, as Lewis puts it, a “favorable attitude”. According to Lewis, a belief is neither a favorable attitude nor does it necessarily inspire favorable attitudes. Hence, beliefs must be rejected as potential candidates for valuation. For if having a belief is constitutive of valuing, “we might not have favored the things we value. We might have opposed them, or been entirely indifferent.”[v]
Lewis seems to be relying here on the Humean distinction between beliefs, which for Hume are strictly non-conative mental states, and desires, which are intrinsically motivating.[vi] If so, then he is certainly correct that desires are better than beliefs as candidates for valuations. However, it’s not altogether clear that such a strict Humean reading is warranted when we consider beliefs about desires. Michael Smith, for instance, contends that valuing “is a matter of believing [what] we would want to want” under conditions of full rationality, and argues that such beliefs have the requisite favorable attitude.[vii] It’s more likely that Lewis is so dismissive of beliefs for another reason that he mentions; the “circularity” that he worries will prohibit a non-reductive definition. As he says, “we have that being a value is some property such that something has it iff we are disposed, under ideal conditions, to believe…[and therefore] to be right about whether something has it” (emphasis mine) [viii]. The evaluative term on the right-hand side indeed yields a non-reductive definition. “That is not empty”, he says, “but it tells us little.”[ix] This isn’t the best tactic, considering that he never argues for reductive naturalism, but merely states it as an advantage of his broader dispositional theory. This issue falls outside the scope of this essay. Suffice it to say that the fact that Lewis is especially preoccupied in his later explanation of ideal conditions to make sure that they will yield genuine values, and the corresponding criticisms that they might not, is an indicator that perhaps one must build evaluative terms into the analysans. Stronger argument is needed.
Setting that issue aside, desires themselves are problematic candidates for valuing. We are often overwhelmed by or entirely indifferent to our desires in a way that does not seem to apply to values. In the grips of a drug addiction, a man may compulsively desire intoxication but value sobriety.[x] So too a severely depressed person may fail to have or to act on desires at all. Yet it would seem as if such a person has values. A more common example is simply an everyday version of weakness of will. I value my health, and as such attempt to avoid the shelf where the dark chocolate is kept. But sometimes my willpower is not enough. If desires themselves were values, such tensions could presumably be chalked up to evaluative conflicts; the addict values the drug more than he values sobriety, the depressed person values her sedentary malaise more than anything else, I value the dark chocolate more than my health, and so on. But this doesn’t seem to be the case. Instead, the individuals in these examples seem to be valuing contrary to their desires. Any theory of valuing is going to have to deal with such evaluative defeasibility.
Lewis’s answer, as already noted, is that desires themselves are not constitutive of valuations. Rather, second-order desiring constitutes valuing. This answer effectively solves the favorable attitude and defeasibility problems presented so far. Since desires are motivating, they a fortiori carry with them a favorable attitude. Also, in each of the aforementioned cases, the second-order desire is identical to the value cited. The addict first-order desires the drug, but second-order desires (values) sobriety. The depressed person lacks a first order desire, but is likely to desire to desire (value) vitality. I desire chocolate all the time, but desire to desire (value) it only occasionally. And in each case, the value may not be satisfied because of a conflict between desires. Just as a desire to wake up early may not be strong enough to overcome a desire to sleep-in, so too may a second-order desire to desire waking up early not be strong enough. So far, Lewis’s definition of valuing is a success.
Problems with higher-order desires
Unfortunately, Lewis’s quick reference to second-order desires engenders just as many problems as he purports to solve. For instance, once we posit a second-order desire, what stops a regress? I desire to desire health, but perhaps I also desire to desire to desire gluttony. Where does the regress stop? Lewis’s answer is not very helpful. He says, rightfully I think, that if we are to adopt higher-order desires as constitutive of valuing, then we need to reserve the highest-order value as a place-holder for those times when we may desire to value differently. But the highest-order desire could be at any level. So at what level do we stop between the penultimate-order desire and the second-order desire? Lewis claims that “it is hard to imagine proper test cases”[xi] beyond the second-order because higher-order desires are meant to be intrinsic rather than instrumental.[xii] The intrinsicalness requirement is puzzling. It may help stop the regress by way of conceptual dumbfounding (i.e. “it’s hard to imagine”), but at what cost? After all, many of our values seem instrumental. The addict who values sobriety may do so for its own sake – to match some personal ideal – or he may simply see it as a means to please his parole officer. The depressed individual may value mental vitality merely as a means to secure a good job. I might value my health as a means to stay attractive into old age. Likewise, it may be that second-order desires can’t be intrinsic at all by dint that they always have as their objects other desires. A good will may be an intrinsic good. But is a desire to desire a good will intrinsically good? We need not belabor such a puzzle to see that in the absence of some special argument the intrinsicalness requirement is simply too problematic to maintain. It’s best, then, to amend Lewis’s definition of valuing to include instrumental second-order desires, and try to solve the regress problem in another way.
One potential solution is the substantive claim that, as a matter of empirical fact, human beings are the sorts of creatures who are complex enough to have second-order desires, but are too simplistic to easily surpass the second-order. This would be enough to stop the regress in this world. And perhaps that’s all that a higher-order desire theorist needs. The claim is initially plausible for at least two reasons. The first is anecdotal, though not insubstantial. During discussions of evaluative problems we often hear such phrases as “I wish I didn’t care so much about my appearance”, “I’d like to have the desire to wake up early”, or “I enjoy my afternoon whiskey and I wouldn’t have it any other way.” Each of these phrases indicates a second order desire. (In fact, each phrase indicates a different sort of higher-order desire to be discussed shortly.) However, it’s rare to hear such phrases as, “I desire to wish that I didn’t care so much about my appearance”, “I don’t want to desire to want to wake up early”, and so on. The second, more principled reason is that a theory ought to be no more complex than necessary. Considering that second-order desires already multiply complexity, positing desires-as-evaluations beyond the second-order seems to be an extravagance.
Unfortunately, a priori considerations substantially diminish the force of the idea that human beings can’t actually third-order desire. Contra-Lewis, it’s not that hard to imagine test cases. The strongest version of the claim is that there is no way for humans to hold third-order desires at all. This claim is easily defeated; at this very moment I hold a desire for the satisfaction of a desire that I desire to desire. Such desiring is trivial, of course, as it is parasitic on the lower-order desire which is its object whenever there is no countervailing reason for me not to desire to desire to desire. Iterating the question – do you (n-order) desire your (n-1 order) desire? – might yield an infinity of desires whenever this is the case. This is enough to defeat the claim that we can’t third-order desire. But it is a hollow victory. These higher-order desires are explanatorily extraneous. They explain neither attitude nor action any more than the original desire on which they are parasitic. Furthermore, if such desires constitute valuations, they are a sort of mirror valuation, capable of giving us no more than the original.[xiii] A slightly weaker claim is that there is no way for us to hold non-trivial third-order desires. But this isn’t right either. I may come to desire the dissatisfaction of a few of my second-order desires. My family may have raised me to desire only pious desires. But I don’t. Instead, I compulsively desire a few impious desires. My regular church attendance may eventually inspire me to desire only desires for pious desires. Hence, I can hold a non-trivial third-order desire. Notice too that such a desire seems to constitute a valuation when we consider the compulsiveness and conflict of the second-order desires. That is, the third-order desire is not theoretically extraneous.
The trouble is that the sort of empirical evidence that we would need to corroborate the claim that humans actually can’t third-order desire is not readily available. And since there is no a priori argument establishing that third-order desires are everywhere impossible for humans, and it is conceivable that at least some humans have non-trivial third-order desires, third-order desire advocates have a salient response. Conceivability entails possibility; humans can possibly non-trivially third-order desire.[xiv] In the absence of contrary evidence, there’s nothing to show that the possibility of humans non-trivially desiring beyond the second-order doesn’t obtain in this world. The empirical claim is not viable in stopping the regress at any definite level.
However, the empirical claim is not without theoretical worth. There is obviously some threshold to how high we can desire. No one would sincerely dispute the weaker empirical claim that humans in this world cannot hold a non-trivial 10th-order desire. Hence, positing desires beyond the 10th-order is ontologically extravagant; such a high-level of desire couldn’t explain anything that a human can do in this world. Hopefully, psychology will have something to say about where our threshold is. Suppose for the moment that humans in this world can (perhaps with a bit of practice and a fresh pot of coffee) hold non-trivial desires up to the 9th-order. Intuitively, non-trivial desires will become increasingly challenging to hold as they approach the threshold. We may liken the holding of non-trivial higher-order desires to other difficult mental tasks; mathematical calculations, memorization, etc. We should, then, revise the original empirical claim by saying that desires beyond the second-order increasingly diminish in cognitive stability. (I return to this issue in section II.)
This empirical claim, if true, will help explain why a value winds up residing at one order or another. But it does not give us a principle to know exactly where in the desire hierarchy a value resides. Saying that values tend to hover around the lower-order desires due to increasing cognitive instability is not much better than Lewis’s answer to the regress problem, which as Michael Smith puts it, “is simply arbitrary [in that] he could equally well have chosen any level other than the first or highest…it follows that we cannot identify valuing with desiring to desire at any level.”[xv] Smith also criticizes Lewis for a similar sort of arbitrariness in the harmonizing between first and second order desires. The drug addict desires to desire sobriety. But he desires euphoria. The two desires can harmonize one way or the other. He could reconcile his first-order desire with his second-order desire by desiring sobriety, or by desiring to desire euphoria. According to Smith, “there is simply no reason to assume that reason is on the side of achieving that harmony by changing our first-order desires to suit our second-order desires rather than vice-versa. On the Humean’s maximizing conception of rationality it all depends on which desire is stronger.”[xvi] But intuitively, reason is on the side of desiring what we value. Both of Smith’s criticisms are aimed at pushing Lewis’s account to require an irreducible rational standard in the analysandum.
II. Modifications to higher-order desire theory
Neither criticism is insurmountable. To see why, consider three modifications to Lewis’s theory. One modification is to recognize an ambiguity in the contemporary definition of second-order desires. A second order desire is “a desire to desire.” But does this mean a desire for the existence of a first-order desire, or does it mean a desire for the satisfaction of a first-order desire? I take both Lewis’s explanation, Frankfurt’s original paper on the matter and Smith’s criticism to be favoring the former interpretation. But even this first interpretation is also ambiguous. We could mean a desire for the existence of a desire or a desire for the non-existence of a desire. Drawing these distinctions is crucial to countering Smith’s criticisms, and a few more besides.
The first kind of higher-order desire is the desire to have a desire. Call these existential desires. The depressed individual may have existential desires, though she lacks first-order desires. That is, she desires some first-order desire to exist. Existential desires can be negative as well. Russell pointed out that we may “desire not to have…those (desires) which interfere with our natural passions.”[xvii] For instance, a theist may desire nothing more than convincing proof of the existence of god – perhaps the occurrence of clearly identifiable miracles. When such proof consistently fails to obtain, the theist may form a negative existential desire. She may desire to not desire the convincing proof of god, lest she lose her faith. Notice though, that negative existential desires constitute disvalues. The theist no longer values the convincing proof of the existence of god. Positive existential desires best explain scenarios where there is a dearth of first-order desiring (as in the depression case). And negative existential desires best explain cases in which first-order desires are frustrated by the world (as in the theist case).
However, existential desires do not so clearly explain the traditional cases that Lewis considers. It is supposed to be clear that the drug addict first-order desires the drug, but second-order desires to have a desire for sobriety. Mightn’t the addict already have this desire, or something similar? He desires the drug. But he is also likely to desire the company of his former friends, his parents’ admiration, a steady job, and other such things as well as sobriety, all of which conflict with his desire for euphoria. This leads us to a different type of higher-order desire. It is a higher-order desire for the satisfaction of an existing desire. Call these satisfaction desires. Satisfaction desires are almost always adjudicating. For instance, I currently desire to workout at the gym. But I also desire to finish drafting this section of the paper before the weekend. The two desires conflict. In this case, a second-order satisfaction desire adjudicates between the two – it indicates which one I desire to be satisfied[xviii] and which one I desire not to be satisfied. In this case, I desire that my scholarly desire be satisfied and my athletic desire not be satisfied. The adjudicating desire need not, of course, have any causal influence on which first-order desire is actually satisfied, nor does it necessarily include any sort of conscious, willful judgment or rationale. (On my sense of the term, a roll of the dice or a game of rock-paper-scissors can adjudicate a conflict.)
Existential desires are desires in the narrow sense that excludes a continued appreciation for what one has. Satisfaction desires are desires in the wide sense that includes a continued appreciation for what one has; i.e. the desire that an existing desire should continue to exist. Satisfaction desires do not disappear when they are satisfied. But they also cannot appear unless they have an existing first-order desire as their object. Existential desires and satisfaction desires often overlap, but are distinct. Take an everyday case. The miserably rich businessman who sees the contentment of his blue-collar workers may desire that he himself have a desire for modest labor. That is, he has a positive existential desire for the desire for modest labor. The miserably rich businessman’s desire for the existence of a desire for modest labor may be satisfied without him in turn desiring the satisfaction of his newly-formed desire for modest labor. He may still shun modest labor when confronted with extravagance. Likewise, in the case of the drug-addict, his desire for euphoria wins-out every time against his desire for sobriety. His satisfaction desire, then, is an adjudicating desire. It is one that desires the satisfaction of an already existing but unfulfilled desire.[xix] But the drug addict may not have a positive existential desire for any desires that he does not already have, or a negative existential desire for the deletion of an existing desire.
This difference between existential and satisfaction desires provides some of the materials needed to solve the problems in Lewis’s theory. In particular, it is sufficient to show that Smith’s harmony criticism is not as substantial as it at first appears.
Response to Smith’s harmony criticism
Smith’s harmony criticism consists of two premises. The first premise is that accepting a normative reason to Φ is the same as valuing Φ, and that if we accept that we have normative reason to Φ, then we rationally should desire to Φ.[xx] In other words, someone who accepts that she values Φ “and yet, desires not to Φ, should get rid of her desire to not Φ and acquire the desire to Φ instead.”[xxi] The second premise is that “on the Humean’s maximizing conception of rationality (what it is rational to do) all depends on which desire is stronger.”[xxii] The conclusion, as previously mentioned, is that if a valuation consists in desiring to desire, then the two premises are inconsistent as stated. “There is simply no reason to assume that reason is on the side of achieving that harmony by changing our first-order desires to suit our second-order desires rather than vice-versa.”
Suppose we assume for the moment that the particular conception of rationality in each premise is correct. Notice that Smith appeals to existential desires – not satisfaction desires. He writes of “acquiring” and “getting rid of” desires. Satisfaction desires already consistently meet the rational standard in either premise. The reason is that a satisfaction desire already has as its object an existing first-order desire. So if we accept a valuation consisting of a satisfaction desire to desire Φ, we cannot fail to desire Φ. The harmony criticism, then, does not apply to a substantial portion of valuations.
I will argue that the harmony criticism only applies when 1. there is a valuation that does not have a harmonious first-order desire (existential desires) and 2. there are no harmonious competing valuations (no satisfaction desires) and 3. there are first-order desires. To see that this is the case, consider first our miserably wealthy businessman. He has a positive existential desire to have a desire for modest labor. By definition, he does not yet have this desire. So it would seem that the harmony criticism might apply; there is at the very least a disharmony between his positive existential desire to Φ and his lack of a desire to Φ. However, if he has even one satisfaction desire, then it follows that he values something other than, and therefore has a reason not to desire, modest labor.[xxiii] Again, the harmony criticism does not apply. He will not be in violation of the rational criteria in either premise. Conditions 1 and 3 are met, but not 2. It would have to be the case that the wealthy businessman’s only valuations consist of positive existential desires. This is not inconceivable. But it’s unlikely for functional human beings. I’ll return to this possibility in a moment.
The depressive case is a much more common example of positive existential desire. The depressive has no satisfaction desires because she has no first-order desires. She has only positive existential desires. Nevertheless, the depressive’s reasons are not in violation of the rational standards. Given that she does not have any first-order desire, it is impossible (on the Humean maximization conception of rationality) for her to have conflicting reasons. She most desires the existence of first-order desires. For specificity, suppose she values (desires to have the desire for) anti-depressent medication. According to Smith’s first premise she now has a rational reason to desire medication. But because there is no conflicting first-order desire, it is clear, according to Smith’s second premise, that she has rational reason to do what she most desires to do. Again, she most desires to desire medication. So she has a rational reason to desire medication. Smith’s harmony criticism does not apply to depressives. Conditions 1 and 2 are met, but not 3.
Consider next negative existential desires. The theist wishes she didn’t have a desire for convincing proof of the existence of god. She disvalues such proof. Our retort is not dissimilar from the case of the businessman. If the theist has even one satisfaction desire for something other than the convincing proof of the existence of god, then it follows that she values something other than, and therefore already has a reason not to desire, convincing proof of the existence of god.[xxiv] Conditions 1 and 3 are met, but not 2. Like the businessman, in order for the harmony criticism to apply, her only values will have to consist of negative existential desires.
If our distinctions regarding second-order desires are correct, then Smith’s use of the old paradigm of desiring to (have) desire has made his harmony criticism apply to a severely limited class of valuations. So far, it seems that the only valuers that Smith’s harmony criticism applies to are those who have first-order desires that are disharmonious with valuations consisting only of existential desires. Hence, Smith’s harmony criticism is not as substantial as it at first appeared. This is no small accomplishment, considering that we have simply taken it for granted that the rational standards in his premises are true and appealed to nothing more than the intuitive distinction between satisfaction and existential desires. Admittedly, the arguments in this paper are not enough to totally eliminate Smith’s harmony criticism.[xxv] However, when put in the context of a broader dispositional theory of value, a higher-order desire theory need not accept both of the rational standards tout court. Ideal conditions (however that’s spelled out) may alter the strengths in a system of desires according to the content of those desires. So if we are attracted to the idea that what really constitutes value is what we would higher-order desire in such-and-such conditions, then the Humean conception may need revision. Likewise, there may be cases, such as the serendipitous satisfactions that I mention later in this section, when we have rational reason not to desire what we value, in which case the idea that a value is the same as a normative reason does not entail that we have a reason to desire something.
Second modification
The next modification is a principle to explain where in the evaluative hierarchy a value resides. This is where Frankfurt comes in. He claims that there won’t be a third-order desire unless there is a conflict at the second-order. Again, this follows from the modification already made. Higher-order satisfaction desires best explain conflicts at the predecessor-level of desire and higher-order existential desires best explain frustration or lack of a predecessor-level desire. Coupled with Frankfurt’s claim that higher-order desires arise because of turmoil of lower-order desires, we have a non-arbitrary explanation as to what place constitutes a valuation in the potential regress of desires.
(D1) An existing n-order desire constitutes a valuation (where n is a positive natural number) iff there is a conflict, frustration or lack of n-1 order desires and there is no conflict or frustration of n-order desires.
We ought to pause here to see that already (D1) counters Smith’s arbitrariness accusation because there is now a principled means of deciding which order of desire constitutes a valuation. There is yet another modification to be made. Before continuing, we need to consider a potential problem with (D1) and the distinction between second-order desires.
If second-order desires arise from first-order conflicts, there’s nothing in principle to keep similar conflicts from occurring at the third order, and the fourth, fifth, and so on. Suppose a subject, call her ‘Sue’ has a conflict between second-order desires. She might, for example, desire to desire fine cuisine and desire to desire financial prudence. A third-order desire may emerge to adjudicate the value, but then conflicts may occur at that level. One interesting consequence of the distinction between existential and satisfaction desires, is that existential desires may come about because of satisfaction desires. For instance, Sue may form a desire for her desire for financial prudence to be satisfied. Unfortunately, this satisfaction desire is always frustrated because she spends copious amounts of money on fine cuisine. This in turn may bring forth a third-order existential desire. This third-order existential desire may be negative in that it is a desire to not have the satisfaction desire, or it may be positive in that it is a desire to have a different sort of satisfaction desire. There are two potential regresses here. 1. If the third-order existential desire is negative, and it is satisfied, then we run the risk of recursion (a kind of regress). Conflicts at the first level bring about second-order satisfaction desires, which when frustrated bring about third-order negative existential desires, which when satisfied delete the second-order satisfaction desire, which leaves the first-order conflicts unresolved. The unresolved first-order conflicts again bring about satisfaction desires…and so on. 2. If the third order existential desire is positive, then we create a second satisfaction desire that may compete with the first, running the risk of a regress of higher-level conflicts. Hence, we not only again run the risk of regressing throughout the hierarchy of desires, but now also run the risk of endless recursion. We’ve solved one regress at the cost of two more!
Fortunately, the regresses are not destructive. Notice that the recursion case can easily end if the first-order desires no longer conflict, or if a different satisfaction desire emerges. The regress case can end if any one of the higher-order desires in the regress does not conflict with a desire of the same order. So the regresses are not necessarily destructive. And even in those cases where they are destructive, they may not be destructive to the theory so much as the valuation. In fact, the destructive regresses may actually help the theory. There are three possible outcomes of such turmoil. As Frankfurt tells us, one outcome is simply that, “if [the turmoil] is so severe that it prevents him from identifying himself in a sufficiently decisive way with any of his conflicting first-order desires, destroys him as a person.”[xxvi] We might retain the spirit but not the content of Frankfurt’s original claim (he was after all writing about freedom of the will and personhood) and conclude less drastically that regressing beyond one’s own threshold or an infinite recursion destroys the ability to value (which in turn may destroy you as a person). This is a plausible explanation for a few extreme cases. Think of aboulomania, the disorder characterized by pathological indecisiveness. Think again of a manic-depressive in the grips of a mania. If higher-order desires fail to adjudicate, she may try in vain to satisfy all sorts of desires simultaneously, unfortunately accomplishing nothing more than a hodgepodge of half-deeds. At least part of what defines such maniacal cases is an inability to truly value anything.
The other two possible outcomes of evaluative turmoil occur “when a person identifies himself decisively with one of his first-order desires.” As Frankfurt says, “this commitment ‘resounds’ throughout the…array of higher-orders.”[xxvii] This is where our revised empirical claim comes in handy. If it is true that it becomes increasingly difficult to hold non-trivial desires as we approach the threshold, then a fortiori it is true that it will be difficult to hold conflicting or frustrated higher-order desires as we approach the threshold. If so, then it is likely to be the case that there will be increasing cognitive pressure to resolve conflicts one way or the other as we ascend the hierarchy of desires.
The third modification
Frankfurt tells us, rightfully I think, that conflicts produce higher-order desires. But those higher-order desires may be so cognitively unstable as we approach the threshold that we are unable to hold them for very long, if at all, leaving us with conflicts or frustrations at a lower-level that we will either have to resolve, or simply live with. Consider the former. We often have severe inner turmoil, perhaps so much that we lose sight of exactly what we value. But eventually the inner turmoil resolves itself when we realize (or decide) that we value one thing and not the other. The miserably rich businessman who is served divorce papers may realize, all too late, that he didn’t value his money after all. The conflict is, then, dissolved when we desire as we desire to desire. All this is fully consistent with (D1). However, what if the miserably rich businessman discovers that he has conflicting valuations that he must live with? This latter option – simply “living with” conflicts – is a bit trickier.
I value my life. I value my liberty. There may come a time, especially in this turbulent political atmosphere, when I can’t have both. The last few sections make it clear that (D1), together with the empirical claim that higher-order conflicts are cognitively unstable, puts increasing pressure on us to resolve such conflicts as we ascend the desire hierarchy. As Frankfurt says, if we do not identify ourselves in a sufficiently decisive way with any of our conflicting first-order desires, then we lose our ability to value. Does this mean that, if the time comes, I simply can’t value both my life and my liberty? The idea is absurd. During the pre-WWII rise of fascism those who chose their life over their liberty certainly still valued liberty, and those who chose their liberty over their life certainly valued, up until their untimely demise, their life. However, there is a subtle difference between resolving a conflict by identifying ourselves decisively with a desire and dissolving a conflict by doing away with any conflict at all. The latter option is drastic and in a variety of scenarios seemingly impossible. Part of the weltschmerz of everyday decision making is that we can’t have it all (though we usually want to). Resolving a conflict or frustration need only mean decisively embracing one higher-order desire over another. The conflicting or frustrated desire can remain, and perhaps continue to cause anguish. But it need not destroy our ability to value. Resolvable conflicts are, then, those conflicts involving desires which do not have to generate higher-order desires in order for us to choose one or another of them (most likely based on the strength of the desire). Irresolvable conflicts are those that must generate desires of the higher-order. (D1) now becomes:
(D1*) An existing n-order desire constitutes a valuation (where n is a positive natural number) iff there is an irresolvable conflict, frustration or lack of n-1 order desires and there is no irresolvable conflict or frustration of n-order desires.
There is at least one further criticism that (D1*) can counter. Mark Johnston mentions “cases in which one values and yet has reason not to desire, and hence not to desire to desire, reasons which might be thoroughly effective.”[xxviii] The example that he gives is of serendipitous satisfactions.
…satisfactions which as a matter of psychological fact or artificial circumstance come only to those who are without any prior desire for them. To those who want such satisfaction they shall not be given. Knowing that some satisfaction was strongly serendipitous in this sense, I could reasonably and without doublemindedness value it and precisely not want to want it. Notice that this kind of point could be made at any place in the hierarchy of higher-order desires, throwing into question the identification of valuing with any order of iterated desire. Such considerations suggest that the attitude of valuing may be none other than the attitude of judging valuable.[xxix]
As an ad hominem, it’s worth mentioning that this argument is apposite in debunking one Johnston gives just two paragraphs prior. The iterativity of value states that, “Among the things that are valuable is: to value the valuable.”[xxx] But of course, just as there might be times when one has reasons not to desire to desire, one might have reasons not to value the valuable. Consider the case of the spiteful boss. A promotion at work is good for you. But if you actually value a promotion at work, the spiteful boss won’t allow it to happen. You now have reason not to value a promotion.[xxxi] Hence, Johnston’s serendipitous satisfactions show that the iterativity of value is not a priori, as he wishes it to be.
Aside from this attack against Johnston’s consistency, Johnston is hasty to assume that because we have a reason not to desire, we have a reason not to desire to desire. In fact, if the second-order desire is one which is a desire to rid oneself of a – or even all – first order desire (i.e. a second-order negative existential desire), then one can effectively not desire while still having a value. The solution works for any point in the higher-order hierarchy. If an n-1-order satisfaction desire is serendipitous, all one needs is an n-order negative existential desire to be satisfied. That way, just like depressed individuals, s/he can have a valuation without having lower-order desires.
These modifications to Lewis’s theory only provide the appropriate favorable attitude that constitutes valuing. It presently remains a mystery as to why higher-order desires arise and furthermore why they identify decisively with one lower-order desire rather than another.[xxxii] My story, like Lewis’s, is causal, but the discussion falls outside of the scope of this essay.
III. A lingering worry about the modifications
(D1*) allows an existing first-order desire to constitute valuing on those rare occasions when there are no conflicts or frustrations. There is much to say about this particular issue. The biconditional can be modified so that nothing lower than the second-order of desiring can constitute valuing. In a moment I provide such a modification. But it would be hasty to do so without first reevaluating the possibility that, contrary to our previous dismissal, first-order desires might not be such bad candidates for valuing. The original criticisms against first-order desires no longer apply. The drug-addict’s first-order desires cannot constitute values because those desires are conflicted and frustrated. And (D1*) tells us that such desires cannot constitute evaluations. So the drug addict will have higher-order satisfaction or negative existential desires. He desires the satisfaction of his desire for sobriety or the non-existence of his desire for euphoria (or most likely both.) Similarly, the depressive indeed values, despite her lack of first-order desire because, as (D1*) tells us, her lack of first-order desire is exactly what generates the positive existential desires that now constitute her valuing. As long as there is frustration or conflict, the idea that first-order desires can constitute valuations never comes into question. But it is not inconceivable that first-order desires may be without conflict or frustration, and hence themselves be valuations.
Imagine a desire utopia in which one has no conflicting first-order desires because every first-order desire that one has is satisfied from birth till death. In such a world, is it too theoretically frugal to assume that second-order desires would not emerge? I think not. Let’s return to the case of the drug addict. He desires the drug. But he is also likely to desire the company of his former friends, his parents’ admiration, a steady job, and other such things, all of which have been frustrated by his desire for euphoria. Now suppose that all of those desires were magically satisfied without conflict. Is there any need for the drug addict to have desires for the existence or satisfaction of his desires? It seems that he does not. The question is now whether any of his first-order desires can constitute valuing.
The suggestion that desires simpliciter can constitute valuations may illicit the response that even the dullest of animals can have first-order desires. Yet they don’t seem to value anything. Frankfurt goes so far as to state that second-order desires are a requisite of personhood, referring to those humans that do not second-order desire as “wantons”, inferior to persons in that they have no will of their own. Likewise, it’s intuitive (at least to me) that even if first-order desires can constitute valuations, not all of them can, even in this highly idealized scenario. This indicates that there must be some criterion to distinguish between first-order desires that are values and those that are just valueless desires. But the value-as-desire theorist can absorb these criticisms. Frankfurt is probably correct in saying that second-order desires are a requisite of personhood. There is no reason for the desire-as-value theorist to deny this. But the claim that second-order desires are a requisite of personhood does not entail that higher-order desiring is a requisite of valuing. Perhaps the drug addict whose first-order desires are magically satisfied is valuing, but nevertheless he is not yet a person.[xxxiii] For ease of reference, call a human who has never had any conflict or frustration of first-order desires a ‘utopian’.
At first glance the utopian seems to be on a par with wantons and non-human animals. But this is not the case, for a certain counterfactual holds in the case of the utopian that does not hold for wantons and non-human animals. If the utopian’s first-order desires were to conflict or be frustrated, second-order desires would surely emerge. And it is only those desires that he would desire (perhaps under certain conditions[xxxiv]) to either exist or be satisfied that constitute valuing. Hence, even in the highly idealized utopian scenario, only some first-order desiring constitutes valuing. The suggestion gains further plausibility when we consider that such pure satisfaction is incredibly rare. In this world the frustration and conflict of desires is one of the first things to happen to us. As a result, we are certain to have plenty of second-order desires.
We must now amend (D1*). The general form of the conditional for determining whether or not a first-order desire constitutes valuing is: an existing satisfied, non-conflicted first-order desire constitutes valuing just in case, if it were to conflict, then a satisfaction second-order desire would emerge to reinforce it. Added to (D1), we get:
(D1**) An existing n-order desire constitutes a valuation (where n is a positive natural number) iff there is an irresolvable conflict, frustration or lack of n-1 order desires and there is no irresolvable conflict or frustration of n-order desires and, if n=1 and n-order desires were to conflict, then a satisfaction second-order desire would emerge to reinforce it.
(D1**) should be enough to counter the criticism that (D1*) somehow equates human persons with non-human animals and wantons. All and only persons satisfy (D1**). Wantons and non-human animals cannot value because when their desires conflict or are frustrated, no higher-order desires emerge. The account is dispositional in regards to second-order desires (though not those desires that are higher than the second-order).
But if the idea that in certain highly unlikely scenarios at least some first-order desires can constitute valuations is contentious, a theorist attempting to reduce valuation to desires need not take on the theoretical baggage. Such scenarios are incredibly unlikely to occur anyway. So why quibble? The higher-order desire theorist can offer an amended bi-conditional.
(D2) An n-order desire constitutes a valuation (where n is a positive natural number greater than 1) iff there is a conflict or frustration of n-1 order desires and there is no conflict or frustration of n-order desires.
On this definition, it is never the case that first-order desires constitute valuing, because the lowest possible numerical value in the analysandum will be 2, and therefore 1 in the analysans. The higher-order desire theorist must, of course, give a non-stipulative story about the restrictions on n. The story has already been given for (D1**). For (D2) the story will be a bit more complex, though as noted the intuition is already well-established in the literature that second-order desires are the lowest order that can constitute valuing. I’ve provided reasons why this intuition may be incorrect. But there is no need to belabor the point any further. Employing either biconditional will be a vast improvement on traditional reductions of valuing to desiring. Either (D1**) or (D2) counters the criticisms by Johnston and Smith, and probably a few more besides.
(D1**) An existing n-order desire constitutes a valuation (where n is a positive natural number) iff there is an irresolvable conflict, frustration or lack of n-1 order desires and there is no irresolvable conflict or frustration of n-order desires and, if n=1 and n-order desires were to conflict, then a satisfaction second-order desire would emerge to reinforce it.
(D2) An n-order desire constitutes a valuation (where n is a positive natural number greater than 1) iff there is an irresolvable conflict or frustration of n-1 order desires and there is no irresolvable conflict or frustration of n-order desires.
We have, then, a viable solution to the evaluative turmoil of regressions, conflicts and frustrations. Embrace them. They help explain the troubles encountered in even the most mundane instances of valuing, and therefore bolster rather than undermine a higher-order desire theory.
[i] Donald Davidson, “Intending” in The Essential Davidson, Oxford University Press, 2006, pg 124-125.
[ii] My knowledge of Gauthier is mostly second-hand from Michael Smith, The Moral Problem, Blackwell Publishing, 1994. Pgs 141-142. However, a first read of Gauthier shows that he quite explicitly endorses the value-as-desire reduction. “Desire, not thought, and volition, not cognition, are the springs of good and evil,” he says. See David Gauthier, Morals by Agreement, Clarendon Press, Oxford, 1986 pg21.
[iii] Bertrand Russell, “Is Ethics a Branch of Empirical Psychology” in Russell on Ethics, ed. Charles Pidgen, Rutledge 1999, pgs71-78. Also see Michael K. Potter, Bertrand Russell’s Ethics, Continuum, London, 2006, pgs 3-4. Apparently, Russell’s advocacy impressed Moore enough to target it in his Principia Ethica.
[iv] Lewis Ibid pg. 116
[v] Lewis Ibid pg. 115
[vi] See David Lewis, Desire as Belief, Mind, New Series, Vol. 97, No. 387 (Jul., 1988), pp. 323-332
[vii] Smith Ibid. Pgs 150 and 181
[viii] Lewis Ibid pg. 114
[ix] Ibid
[x] The scenario is from Harry Frankfurt, Freedom of the Will and the Concept of a Person, The Journal of Philosophy, Vol. 68, No. 1 (Jan. 14, 1971), pg.9
[xi] Lewis Ibid pg. 155
[xii] Lewis Ibid ftn 4.
[xiii] This is in anticipation of Mark Johnston’s ‘Iterativity of Value’ criticism Johnston ibid pgs 158-159. Here I’ve shown that even if it were the case that we ‘value the valuable’, such valuations are trivial. In section III I show, using Johnston’s own example, that to value the valuable isn’t even a priori.
[xiv] And it does no good to claim that human-like beings who can hold non-trivial desires higher than the second-order are somehow superhuman or superpersons. Suppose that a good friend calls you up in distress and without a hint of philosophical pedantry tells you about a predicament involving third-order desires. “Obviously I desire to desire a trip to the Bahamas”, she might say. “But when I reflect on it I realize that I actually desire to desire to desire staying home.” In this scenario you might question whether or not your friend really does hold third-order desires. But there doesn’t seem to be any reason to question her humanity or personhood. It seems clear that she is still a human person, regardless of her apparent advanced capacity to third-order desire.
[xv] Smith Ibid pg 147.
[xvi] Smith Ibid pg 145
[xvii] Russell Ibid pg.78
[xviii] In his original paper on the subject Frankfurt says that we can “identify” ourselves decisively with a lower-order desire, implying that there may be some sort of endorsement of an existing desire. But again it is ambiguous whether this is to mean satisfaction of desire or perhaps an existential desire in the wide sense.
[xix] As I’m writing this, I check my Facebook page to see that a friend has posted a status update stating “I wish I had the guts to do the things I want to do.” This is an example of a satisfaction desire. She already has wants. Her second-order desire is not for new wants, but for some number of her current wants to be satisfied.
[xx] This is a paraphrase of his C1 constraint. Smith Ibid pg 143.
[xxi] Smith Ibid 145
[xxii] Smith Ibid pg 145
[xxiii] To be precise, there cannot be a satisfaction desire without an existing first-order desire. And there cannot be a positive existential desire for a desire that already exists. It follows that there cannot be both a positive and satisfaction desire for the same desire.
[xxiv] Unlike positive existential desires, there is nothing other than sanity prohibiting us from having both a negative existential desire and a satisfaction desire for the same desire, though further considerations in the next section explain that the outcome of such irresolvable conflicts is that there can be no value at that level of desire.
[xxv] Smith himself does not accept his second premise. Merely, he argues that the higher-order desire theorist cannot revise the Humean maximization conception of rationality in any non-question-begging way.
[xxvi] Frankfurt Ibid pg16
[xxvii] Frankfurt Ibid
[xxviii] Mark Johnston, Dispositional Theories of Value, Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 63 (1989), pg 160
[xxix] Ibid
[xxx] Ibid pg 158
[xxxi] The example is my attempt to rework a rather tortuous one in Smith ibid pgs 149-150, which originally comes from Parfit.
[xxxii] Frankfurt actually goes so far as to specify that “a person may be capricious and irresponsible in forming his second-order (desires) and give no serious consideration to what is at stake…There is no essential restriction on the kind of basis, if any, upon which (they) are formed.” But again, his purposes were different. Frankfurt Ibid pg 13, footnote 6.
[xxxiii] He is also not a ‘wanton’. Frankfurt means wantons to be indifferent when there is turmoil of first-order desires.
[xxxiv] Determining the appropriate conditions for whether or not the thing being valued is really a value would take us far outside the scope of this essay. Traditional ideal observer theories quibble over whether or not one must be omniscient, omnipercipient, impartial, dispassionate and consistent. See Roderick Firth, Ethical Absolutism and the Ideal Observer, Philosophy and Phenomenological Research, Vol. 12, No. 3 (Mar., 1952), pp. 317-345. Lewis reduces all of this to “imaginative acquaintance”, while Johnston and Smith push for the conditions to be conducive to rationality. Here, I am only concerned with the favorable attitude of valuing, not the values themselves. I have set this issue aside and instead spoken only of those conditions that seem (at least to me) to be reasonable enough to support valuations, not necessarily values.
Posts
