James M. Okapal, Missouri Western State University
I. Introduction
We make choices throughout the day: Which wine goes better with dinner? Do I prefer reading Robert Heinlein’s Stranger in a Strange Land or watching Cinema Paradiso for my evening entertainment? These choices involve comparison. 1 A comparativist says that if a comparison is possible, then the comparison takes place in factors, issues of incommensurability, or other reasons that preclude comparisons.[/note] These properties include values, rights, virtues, obligations, non-evaluative facts, etc.
Ruth Chang has developed the most detailed comparativist view. Her key works include: her “Introduction” to Incommensurability, Incomparability, and Practical Reason, Making Comparisons Count, “Putting Together Morality and Well-Being,” and “All things Considered.” According to Chang, each choice situation is resolved by a single comprehensive value, what she sometimes calls a covering value. 2 This is her Comprehensive Value Approach (CVA) to a theory of rational choice. She contrasts her theory with the Orthodox Approach (OA). A defender of the OA denies the necessity of comprehensive values in resolving a choice. 3 Chang can be read as defending two conclusions: first, that the OA fails to produce an acceptable theory of rational choice; second, that her CVA is a prima facie viable alternative. The arguments used to establish these conclusions are problematic. Before examining these arguments, a better understanding of Chang’s overall comparativist project will be useful.
II. Chang’s Project
One challenge for any comparativist position will be the issues of incomparability and incommensurability. It seems to be a problem for a comparativist theory of rational choice any time the choice involves options so different in terms of the properties they bear that the properties appear to resist comparison. 4 To overcome these concerns Chang relies on two components. The first is the notion of parity and the second is the notion of a comprehensive value. According to Chang, much of the debate concerning incomparability and incommensurability accepts the Trichotomy Thesis that any comparison must include at least one of the following relations: “greater than,” “lesser than,” or “equal to.” 5 Chang argues that the Trichotomy Thesis is false because there is a fourth relation, “on a par.”7 In cases where we find it difficult to see comparisons, this fourth relation helps diminish the possible number of incomparable and incommensurable options, possibly to zero. While this part of her theory is important, there exists a thread in the philosophical literature which discusses the Trichotomy Thesis so it will be left to the side. 6
The second component, comprehensive values, is the focus of this paper. The core idea is that for any given choice situation there exists a single, comprehensive value that indicates a rational choice. In her papers “Putting Together Morality and Well-Being” and “All Things Considered” Chang provides arguments to establish the conclusions that there are problems with OAs and that the CVA should be seen as an alternative. In these articles Chang identifies three issues confronted by a comparativist theory. In “Putting Together Morality and Well-Being” she describes the first issue in terms of features of choice situations. These features should be addressed by a theory of rational choice for the theory to be viable. She argues that current versions of the OA cannot account for these features and her CVA can. In both “Putting Together Morality and Well Being” and “All Things Considered” she addresses “the problem of fragmentation” and “the problem of the unity of value.” The problem of fragmentation concerns the ability for a theory to explain why comprehensive values operate in some choice situations and not in others; the problem of the unity of value concerns the ability of a theory to explain how to unify all the values in a choice situation such that an all-things-considered judgment is possible. Chang concludes that the CVA is plausible because it can address each of these issues while OAs cannot adequately address them.
There are several possible ways to develop a comparativist view. For the current discussion there are two main categories, namely the CVA and the OA. Each category has sub-categories. A defender of a monistic version of the CVA holds that there is a single value of comparison that determines choice in all situations. A pluralistic version, one that Chang defends, concludes that for each conflict “there is some or other more comprehensive value in virtue of which there is a rational resolution” to each conflict. 7 In other words, she accepts that for any two situations x and y, the single value that governs choice in x may be different than the single value that governs choice in y. Furthermore, the nature of this comprehensive value is such that when conflicting values lead to conflicting justifications, the comprehensive value “includes the conflicting values as “parts” and is that in virtue of which the conflict is rationally resolvable.” 8
There exist simple and sophisticated versions of the OA. On simple versions, the values “at stake alone account for their relative weights regardless of the circumstances in which they figure.” 9 She also says that the values “at stake themselves determine their own normative relations.” 10 In other words, each value comes with some prior weighting that already takes into account its normative relation to all other possible values. On sophisticated versions, determining the relative weights of values is a product of the values and supplemental elements. 11 Some versions of the sophisticated OA determine weight by a function of the values and circumstances in which they appear. Other versions determine relative weight by providing supplemental content with a purpose, principle, or theory other than a comprehensive value. 12 The concern of this paper is whether Chang’s arguments show that her pluralistic version of the CVA is a prima facie plausible alternative theory of rational choice and also a theory that has advantages over OAs. The thesis is that none of the arguments considered here turn out to be successful.
Before examining these arguments, it might be helpful to say more about Chang’s terminology. The extension of the terms “comprehensive value approach,” “orthodox approach,” and their subdivisions are unclear. As such, the terminology makes it difficult to identify theories that fall under the concepts. In several footnotes, Chang provides some hints by identifying contemporary theorists and some theory types as exemplars of a variety of OAs. 13 By attempting to identify some views of practical reason and value that could be extensions of the terms should help clarify the concepts
The first step is to realize that a theory of rational choice is composed of two parts. One part is the inferential procedure used to move from stated considerations, such as values and non- evaluative facts, to a conclusion about what one ought to do. Gerald Dworkin in “Theory, Practice, and Moral Reasoning” describes several inferential procedures. For example, the deductive procedure assumes “there are quite abstract principles about act types that, together with the premise that the act at issue falls under one of these principles, determines deductively that the act is permissible or not.” 14 The other part identifies specific considerations to be used in the inferential procedure. Following Betsy Postow, this could be called a Reasons Theory. According to Postow, a “Reasons Theory is a normative theory that purports to tell us what considerations constitute or provide reasons.” 15 In the case of the deductive procedure, the two types of consideration would be the abstract principle and an act description that includes a statement that the act is subsumed under the principle. While every view of rational choice will have both parts, at least implicitly, it is possible to state a Reasons Theory without attaching it to a specific inferential procedure. An example of such a free-floating Reasons Theory includes views that identify desires, either actual or ideal, as the sole reasons we could have to act. Given the way Chang describes CVAs and OAs, the distinguishing marks are found in the Reasons Theory part of a theory of rational choice.
By identifying the supplemental factors as the considerations that provide reasons, specific views can be understood as examples of a CVA or an OA. Consider the variety of sophisticated OAs mentioned by Chang. The most well-known types of OAs are moral theories that identify values and principles as central to their Reasons Theory. Rule-oriented theories such as rule utilitarianism, common morality, and contractualism would count as a sophisticated OA with values and principles. In each case, just knowing which values are at stake is insufficient to determine a justified course of action. These views add principles as tools to organize the values. Once both of these parts are in place, then an inferential procedure can be used to determine the rational course of action. So, traditional act-oriented moral theories that incorporate a deductive inferential procedure would be sophisticated OAs that use both values and principles.
But, as Chang points out, sophisticated OAs do not have to use principles as the organizing tool. First consider a sophisticated OA with values and circumstances as the organizing tools. Despite Chang’s hesitancy, particularism is a very good candidate for such a view. 16 Whatever might turn out to be an example of a fully developed particularist view, such a view will only cite values and non-evaluative circumstances as relevant to rational choice. A fully developed particularist view will have to spell out both how these considerations generate normative relations as well as explain the relevant inferential procedure. Once this is done, however, particularist views seem to be sophisticated OA that identifies only values and circumstances as the core of the Reasons Theory.
One example of a sophisticated OA that Chang does not explicitly mention would include values, prima facie principles, and judgment as part of its Reasons Theory. This is Rossian pluralism. In situations of conflict, this type of pluralist believes that values and principles are insufficient to resolve conflict. An act of judgment is needed which “balances” the principles to determine the appropriate action. This particular view is mentioned for two reasons. First, this author is sympathetic to such a view and, as will become apparent, believes that a view with a strong family resemblance to Rossian pluralism is a major rival to Chang’s CVA. Second, it may be the case that Roassian pluralism as a sophisticated OA and Chang’s CVA are also close relatives in terms of their use of balancing inferential procedures of rational choice. Before exploring either point, however, more work needs to be done identifying the extensions of the other types of OAs and CVAs.
Examples of simple OAs are more difficult to identify. Simple OAs understand values to “alone account for their relative weights regardless of the circumstances in which they figure.” 17 A view that identifies only one type of consideration as truly relevant to practical reason would be part of the extension of the concept “simple OA.” An example of such a view is possibly Donald Hubin’s actual intrinsic motivation view. For Hubin, reasons for acting are ultimately based on ungrounded motivations. These ungrounded motivations are actual intrinsic motivations we have and constitute the values that, once identified, provide “reasons to perform those actions that promote the states of affairs we intrinsically desire.” 18 In other words, the actual intrinsic motivations give us reasons to act, and presumably they would include their relative weights. Given that Chang says so little about these simple OAs, however, identifying Hubin’s view, or any other agent-dependent view of reasons, as an example of a simple OA must be seen as tentative.
This leaves us with trying to identify monistic and pluralistic CVAs. One thing to notice is that CVAs of either type, as comparativist theories of rational choice, will include in their extension quantitative theories of rational choice. Versions of these theories which allow options to be numerically represented by interval scale measurements allow for comparisons in terms of their numerical representations. If one assumes that utility is the only value of note in the decision making process and that all other values are part of utility, this would count as a monistic CVA. Arguably, the utilitarian views espoused by Bentham and Mill would be examples of such views as would versions of decision and game theory that only identify one value in its Reasons Theory. Even though all of the views considered so far as examples of monistic CVAs are quantitative, it is unclear whether the extension will be exhausted by such quantitative theories because it is not clear that a simple CVA would require numerical representations. For example, if the only value relevant in all choice situations is God’s will, then numerical quantification can be avoided.
Chang, as already noted, is trying to develop a pluralistic CVA. Whatever else Chang’s CVA comes to, it should be seen as distinct from the views described above. First, she wants to avoid the view collapsing into some form of the OA. Furthermore, given that the view might endorse a new comprehensive value for every choice situation, it is unclear that she will adopt a quantitative approach to rational choice: some of the values may resist numerical representation. Until she more fully develops her concept of a comprehensive value and adopts a specific inferential procedure, it will remain unclear whether she endorses a purely quantitative theory of rational choice. However, Gerald Dworkin provides a hint that Chang’s CVA might adopt a balancing inferential procedure and that this procedure is quantitative. Recall that Rossian pluralism is an example of a balancing inferential procedure whose final arbiter is an act of judgment, but does not endorse a quantitative view of rational choice. One of the traditional problems with Rossian pluralism is that it leaves undeveloped the nature of a balancing judgment. According to Gerald Dworkin, there is an alternative to Ross’s reliance on intuitive balancing:
A different approach to the task of balancing is to claim that all such weighing must be in terms of values, that is, some feature that is good or bad, or casts favorable or unfavorable light on alternatives being weighed. Such an evaluation may be particularistic in character. In each case of conflict, there is some value that provides a means of evaluating choices. The choice that comes out highest on that value scale is the one that should be chosen. 19
While Dworkin does not cite Chang’s view as the inspiration of this alternative, Chang’s view clearly maps onto his description. By suggesting that the values are put on a scale, this also suggests a quantitative view of rational choice, even though that scale may end up being only ordinal in nature. If we take Chang’s comments in Making Comparisons Count as still part of her considered view on the topic, she seems to adopt a quantitative approach to rational choice. In this book she states that comprehensive values “naturally admit of comparisons with more structure than that allowed by mere ordinality,” and “I cannot think of any nongerrymandered [comprehensive] value that gives rise to merely ordinal comparisons. 20 Thus, she seems to think that all comprehensive values will be able to be represented on an interval scale, even if such representation is in some sense imprecise. 21 Whatever the current details of Chang’s evolving views, a CVA resolves each choice situation by reference to a single value that “includes the conflicting values as ‘parts’ and is that in virtue of which the conflict is rationally resolvable.” 22 In other words, the comprehensive value determines the overall normative relations between the rest of the values. On the supposition that Dworkin’s description of this alternative balancing interpretation of rational choice maps onto Chang’s CVA, we can now partially situate Chang’s view: Chang’s overall theory of rational choice could be one whose Reasons Theory involves comprehensive values and lower-level values which are constitutive of the comprehensive value, and the comprehensive value also plays a role in the inferential procedure by balancing the normative relations of those lower-level values. As Gerald Dworkin points out, however, this approach to rational choice “must be tested by first seeing if some of the choices that we are confident are correct are resolved in that fashion.” 23 The arguments put forth by Chang can be interpreted as attempts to satisfy this challenge. Even though none of the argument that follows requires that Chang be saddled with a balancing inferential procedure, if the criticisms of this paper are on target, Chang’s view would not count as a way to meet successfully Dworkin’s challenge.
III. Features of Choice Situations
The first issue to discuss is that the OA will fail to adequately account for features of choice situations. In “Putting Together Morality and Well-Being” 24 Chang states these features as follows: a theory of rational choice should
(1) include the values at stake, (2) have content beyond those values and circumstances in which they figure, (3) have content beyond a particular weighting of those values in those circumstances and (4) be that in virtue of which there can be genuine disagreement about what the correct weighting of values is. 25
Simple OAs fail because according to the view “the values at stake alone account for their relative weights regardless of the circumstances in which they figure.” 26 If this were true, then when the same values appear in different circumstances, their relative weights would stay the same. However, “[w]hatever the normative weights of these values vis-à-vis one another, their weights are different in different circumstances” and thus the values alone cannot account for these weights. 27 Some sophisticated OAs, on the other hand, take into account the circumstances. According to Chang this move fails because “the values at stake and the circumstances in which they figure underdetermine the way those circumstances affect the normative relations among those values.” 28 Other sophisticated OAs that include additional normative content in the form of principles or purposes still face problems related to the weight assigned values and the ability to account for disagreements.
These arguments point to two possible conclusions. In “All Things Considered” Chang clearly states that her cautious aim “is not to argue that the comprehensive value view must be correct; I only want to say enough in its favor to suggest that, all things considered, it is a view that should be taken seriously.” 29 However, in “Putting Together Morality and Well-Being” Chang states that she “cannot see what could fill this role [of addressing all four features] other than a more comprehensive value.” 30 This is clearly a bolder conclusion than merely establishing the prima facie plausibility of her CVA; it suggests that OAs fail to address adequately these features. Unfortunately, neither the cautious nor the bold conclusion has been adequately supported by her arguments concerning the features of choice situations. In order to support the bold conclusion that OAs do not adequately address these features, we must interpret the features as criteria of theory choice. If this feature-as-criterion interpretation is correct, then Chang must offer independent arguments that establish these features as acceptable before using them to criticize rivals and support her CVA. In order to support the cautious conclusion that CVA is prima facie plausible, a theory is acceptable as long as it has a response to issues identified by the features. If this feature-as-question-identifier interpretation is correct, then the features cannot be used to say that a particular answer to a question is unacceptable—as long as an answer has been given, then the demands of the features have been met. To criticize a given response to a feature requires an argument independent of the features themselves. First, I consider the feature-as-criterion interpretation and show that the arguments for accepting features (2) – (4) are problematic. Second, I consider the feature-as- question-identifier interpretation. On this interpretation, she establishes her cautious conclusion but at the price of establishing other theories as prima facie plausible that should not be considered so.
Of the features articulated by Chang, the first feature is uncontroversial: if a theory of rational choice does not address the values at stake in a choice situation, then the theory would not address the actual content of a rational choice. This is also the one feature that all OAs meet. It is the other three features that are worrisome.
Feature (2), on the feature-as-criterion interpretation, requires that theories of rational choice “have content beyond those values and circumstances in which they figure.” 31 This feature, if it can be defended, would provide a reason to reject simple OAs which say that only values are needed to determine the normative relations between values. If feature (2) is required of theories of rational choice, if a theory of rational choice must have content beyond circumstances and values, then simple OAs must be rejected. But do we have to accept feature (2)?
An argument is needed to accept a feature as a constraint on acceptable theories of rational choice. However, if we use the feature-as-criterion interpretation, Chang’s argument begs the question. First note that feature (2) has actually two components, only one of which is needed to eliminate simple OAs as acceptable theories of rational choice. The two components are (a) that theories of rational choice should have content beyond the relevant values and (b) that theories of rational choice should also have content beyond circumstances. If we do not have reason to accept component (a), then we have no reason to accept the conjunction of (a) and (b). So a new question arises: Does Change provide a reason to accept component (a)? The relevant passage uses an example of choosing between two candidates for a faculty position, Eff and Gee. Eff has technical prowess but Gee has a more enjoyable teaching style.
It makes no sense, however, to ask in the abstract which of two values gives rise to the greater reason. Suppose God is told in a circumstantial vacuum to choose between Eff and Gee with regard to philosophical talent and teaching ability. If there are no specified circumstances, even God cannot know whether Eff’s technical prowess gives rise to a greater reason than Gee’s easygoing teaching style because there is no fact about how those values normatively relate apart from circumstances. 32
Her conclusion is that in the abstract, in other words independent of circumstances, it is nonsense to ask which of two values has more weight. God is incapable of answering this question. The reason she gives for this conclusion, however, is just the conclusion itself: God is stymied “because there is no fact about how those values normatively relate apart from circumstances.” 33 If the “because” clause is the reason to accept the conclusion, this is a textbook example of begging the question. It is nonsense to ask what the weights of values independent of circumstances are only if you assume that there is no fact of the matter. The point of asking is because we need a reason to think there is no fact of the matter, especially since an absolute deontic constraint is exactly the type of thing that has a value independent of circumstances: if torture is always unacceptable, then it is unacceptable regardless of circumstances. If absolute deontic constraints exist, then God surely knows what they are and can, based on their absolute nature, determine that they provide greater reasons than any other. So, Chang has not given us an independent reason to accept component (a) and therefore the argument gives no reason to accept feature (2) as a criterion that eliminates simple OAs as acceptable theories of rational choice. Since the bold conclusion requires that simple OAs be eliminated, she has also failed to support the bold conclusion.
Chang may try to establish feature (2) in a different way. She notes that on the simple view, values must always have the same normative relation to each other regardless of the circumstances. But, as she points out, “this view seems mistaken, for sometimes the very same values have different relative weights in different circumstances.” 34 In other words, if defenders of simple OAs cannot provide an explanation for these changes that are denied by their view, then we should reject such theories of rational choice. Unfortunately for Chang, this argument does not really support feature (2). This burden shifting argument challenges defenders of simple OAs to explain how such changes in weightings can occur since they are disallowed by simple OAs. While this is a challenge to defenders of simple OAs, it is not an argument that feature (2) should be accepted as a criterion of theory choice. On the feature-as-criterion interpretation, feature (2) demands that all theories of rational choice take into account content beyond values and circumstances. The appearance that the weighting of values change in different circumstances cannot establish feature (2) unless the appearance can be shown to be a fact. A defender of simple OAs could have the argumentative resources to deny that this appearance is a fact. Suppose that all values exist in a linear hierarchy and that there exist exhaustive rules for adjudicating between possible conflicts. On this system, all one needs to know is the value and the details of this hierarchy to resolve a rational choice problem. Our epistemic limits may suggest that such a view is not practical, but that is not enough to show that it is an incorrect axiological view. Chang needs to directly show that such an axiological view is wrong before she can utilize feature (2) against theories that suppose such a view. Chang has not provided such arguments and thus this attempt to establish feature (2) falls short as well. Again, this means that she does not have the support she needs for her bold conclusion.
If the status of feature (2) is questionable, then that undercuts any reason to accept features (3) and (4) since they assume the acceptance of feature (2). However, there are independent reasons for rejecting features (3) and (4). According to feature (3), theories of rational choice should have content beyond a particular weighting of values in given circumstances. Again, on the feature-as- criterion interpretation, this feature needs to be defended with an independent argument; again, Chang does not successfully defend this feature.
The argument for feature (3) relies on a distinction between the external and internal circumstances of choice situations. According to Chang, an external circumstance determines “which choice situation one is in, including which values are at stake and which circumstances are internal to the choice situation.” 35 Internal circumstances “play a role in determining the relative weights of the values at stake once the choice situation has been identified.” 36 The other key component is the claim that in a given choice situation, “the values at stake and the circumstances in which they figure underdetermine the way in which those circumstances may affect the relative weights of those values.” 37 Putting these two claims together she generates the following argument:
holding the values at stake and their circumstances in which they figure constant, the values at stake could nevertheless have different relative weights in the very same internal circumstances. This is because circumstances external to a choice situation may determine that what matters in one situation is different from what matters in another, even though in both situations the same values are at stake in the same internal circumstances. 38
The key move in the argument is the claim that relative weights can differ even if we hold values and circumstances constant. She attempts to defend this claim with the following example.
Being physically distant from the victim diminishes the relative weight of one’s duty to save vis-à-vis some competing value only if what matters in the choice situation is something that gives greater weight to doing one’s moral duty than, say, to doing what is supererogatory. If what matters instead is saintliness or doing the most supererogatory act possible, the circumstance of being physically distant would cut the other way. … [Thus the] values at stake and the internal circumstance of the choice situation cannot determine the normative relations among values in those circumstances; something with further content is needed. 39
The problem with this example is that it does not follow the definitions Chang gave for the internal circumstance – external circumstance distinction. Thus, this argument does not justify acceptance of feature (3) as criterion. At first Chang claimed that if you hold “the values at stake and their circumstances in which they figure constant, the values at stake could nevertheless have different relative weights in the very same internal circumstances.” 40 But in the duty-to save example she changes what is at stake. What is at stake in one case is “doing one’s moral duty” while in another case what matters is “doing what is supererogatory.” In other words, the external circumstances about what is at stake changed the internal circumstances such that this change explained the different choices. This suggests that the values and the internal circumstances can be enough to determine the relative normative weights of the values because the external circumstances changed the situation and each situation had its own solution. This is a counterexample to her claim that values and circumstances underdetermine relative weights, not a defense of it. But, as a counterexample, it suggests that feature (3) should not be accepted as a criterion of theory choice. Thus another support for her bold conclusion has faltered.
Finally consider feature (4). According to this feature, a theory of rational choice should have content “in virtue of which there can be genuine disagreement about what the correct weighting of values is.” 41 A genuine disagreement occurs when “once we have agreed on what choice situation we are in—and therefore what matters in the choice—we have agreed on what the correct weighting of the values at stake is. There can be no genuine disagreement within a given choice situation about how the values at stake relate.” 42 In other words, according to Chang a genuine disagreement occurs when two people “agree that a certain principle or purpose applies to a situation and still disagree about which of the two items we should choose according to that principle or purpose.” 43 In her argumentative narrative, she assumes that she has been successful in eliminating views which claim that only values matter and only values and circumstances matter. She now wants to consider sophisticated OAs in which you need values, circumstances, and further content to determine relative normative weight and “this further content is given by a purpose, principle, or theory of value, not by a more comprehensive nameless value.” 44 On this view, “relevant content might be thought to be given by a particular weighting of the values at stake.” 45 But, so her argument goes, this will preclude the possibility of genuine disagreement.
Once we have agreed on what choice situation we are in—and therefore what matters in the choice—we have agreed on what the correct weighting of the values at stake is. There can be no genuine disagreement within a given choice situation about how the values at stake relate; … Since, however, there can be [genuine] disagreement, what matters in choice cannot be understood in terms of particular weighting of the values at stake. 46
There are two missteps in this argument. First, Chang has again misused her own distinction between internal and external circumstances. This argument asserts that by agreeing upon what choice situation we are in, we have also agreed upon the correct weighting of the values for that situation. But how does one determine what choice situation one is in? By consideration of the external circumstances, but all the external circumstances do is determine the values at stake, “they do not, as such, play a role in determining the relative weight of the values at stake once the choice situation has been identified.” 47 So, it is incorrect to say that by agreeing about what choice situation we are in, we have also agreed to a particular set of relative weightings of values. 48 This calls into question the idea that Chang has provided a reason to accept the claim that “Once we have agreed on what choice situation we are in … we have agreed on what the correct weighting of the values at stake is.” 49
The second problem is that she inappropriately saddles sophisticated OAs with the implication that they disallow for genuine disagreement. A genuine disagreement occurs when two people “agree that a certain principle or purpose applies to a situation and still disagree about which of the two items we should choose according to that principle or purpose.” 50 It is not true that once you agree upon what principles are relevant that this is always sufficient to provide determinate weightings and choice. There is still the matter of the application of the principle. Now, some principles, when applied will give determinate weightings such that genuine disagreements are avoided. Consider a duty-to-save example where, on your way to give money to Oxfam, you come across someone who is in need of food. Who should you give your money to, the person who you can see needs the money or the equally needy person on another continent? An argument can be made that the distance between you and the needy persons plays a role articulated by a “physical distance discount principle.” A discount principle is not a value and it is not a circumstance, it is an expression of how a fact—physical distance—reduces the relative weight of a given value—the good of “helping another in need.” Assume that the value of “helping another in need” irrespective of physical distance is fixed at some positive weight w. This does not determine the relative value of helping the two people until we establish the discount rate for physical distance. Suppose that the discount rate for someone you can see is 1.0 and the rate for someone who is not on you continent is 0.1. Thus the weight of buying food for the person in front of you is determined by the function (w • 1.0) while the weight of buying food for the person on the other continent is (w • 0.1) and thus (w • 1.0) > (w • 0.1). In such a case Chang is correct to say that there can be no disagreement about choice.
But not all principles have precise mathematical formulations. Suppose you and your friend have identical wine cellars. Furthermore, you have both been invited to the same dinner party with knowledge of what is being served. Finally, you both use the following principle to determine the appropriate wine to bring to a dinner party.
If the wines of category α are better suited for the entrée than wines of category β, then the suitability concern overrides concerns about quality of particular wines A and B unless the quality of B is so superior to eliminate concerns about suitability of A.
Consider the fact that chicken with a cream sauce is paired, usually, with white wines such as a Chablis. The host is serving csirke paprikás, a traditional Hungarian dish of boiled chicken with a sour cream and paprika sauce served with egg noodles. At first blush, you and your friend think that a Chablis should be brought to the party. However, suppose the two of you also happen to have some Hungarian white wine, Badacsonyi Kéknyelü. In general, a Badacsonyi is more suitable than a Chablis for csirke paprikás due to the spiciness of both the wine and the dish. But, suppose further that the quality of Badacsonyi wines are not consistent from year to year and neither of you have tried this particular year’s Badacsonyi, but know your identical stocks of Chablis to be good. Here we have a situation in which the choice confronting you and your friend is identical but she brings a Chablis believing that the quality of the Chablis to be superior enough to eliminate bringing a Badacsonyi and you bring a Badacsonyi because you do not believe the quality of the Chablis to be enough to eliminate Badacsonyi as a choice. Here we have a genuine disagreement because you and your friend “agree that a certain principle or purpose applies to a situation and still disagree about which of the two items we should choose according to that principle or purpose.” 51 What has happened here is that the principle relies on imprecise notions of suitability and quality as well as the situation involving epistemic deficiencies—you do not know in fact the quality of the Badacsonyi. Even if the quality of the wine was agreed upon, the two of you could apply the principle differently based on the imprecision of the concepts “suitability” and “quality” and thus have a genuine disagreement. Real-life choice situations usually exhibit similar epistemic and precision deficiencies which are sufficient to generate genuine disagreements. But that means Chang’s claims that sophisticated OAs disallow for disagreements is mistaken. This in turn means that she has failed to establish feature (4) as a criterion and failed to establish her bold conclusion.
Suppose that my wine-choice example is not convincing. What is the problem with a theory of rational choice disallowing disagreement? Ronald Dworkin in his “Is There Really No Right Answer in Hard Cases?” argues that in at least the legal realm there is always a single right answer to every question thereby disallowing for any disagreement. Kantians seem to accept this view for all moral and practical conflicts as well. On these views it seems to be a problem for a theory of rational choice if it does not at least reduce genuine disagreements. In an ideal world, a theory of rational choice could be structured so that there is an answer to every question in practical and moral domains, as well as providing ways of adjudicating instances where the practical and the moral realms conflict. This suggests, at first blush, that feature (4) is counter-intuitive with regard to the practical aim of a theory of rational choice, namely, that the theory have a high degree of determinacy. It is hard to imagine that anyone would deny that this is a practical aim even though some, this author included, believe that genuine disagreements cannot be eliminated from the practical and moral realms. This belief entails, in agreement with feature (4), that it is problematic if a theory will make genuine disagreements nonsensical. But to claim that any theory that does not recognize the possibility of genuine disagreement is problematic relies on an assumption about the nature of the practical and moral realms, namely that they do contain irresolvable conflicts. But that means the defense of this particular feature, on the feature-as-criterion interpretation, requires an argument in favor of the view that the practical and moral realms do in fact include conflicts. Chang has not provided such arguments, so there is no reason to accept feature (4) as a criterion. Thus her bold conclusion cannot be supported.
All of the above criticisms rely on the idea that the features should be interpreted as criteria of theory choice which entails that we be given arguments that provide independent reasons for accepting the features. Chang does not provide adequate support for features (2) – (4) interpreted in this way. The alternative is to interpret the features as indicating questions that must be addressed by a theory of rational choice. This interpretation creates a problem for Chang because it limits the use of the features: the features cannot be used to criticize the details of any given theory since a theory merely has to provide responses to these features. On this interpretation, the adequacy of response is not open to direct evaluation but needs independent arguments. If this feature-as-question- identifier interpretation cannot call other views into question, then this interpretation falls short of providing justification for her claim that only a comprehensive value can fill the role demanded by the features. It can still be used to defend CVA as a prima facie acceptable theory, but it can do no more. However, on this interpretation, almost any theory put forward can be considered prima facie acceptable. Such a permissive result questions whether any theory that is initially acceptable under this interpretation really is worthy of attention.
On the feature-as-question-identifier interpretation, the features can be restated as follows:
(1*): What values are at stake in rational choice?
(2*): What content beyond values and circumstances are relevant to rational choice?
(3*): What content beyond a particular weighting of values in given circumstances are relevant to rational choice?
(4*): Can there be genuine disagreements and if so, in virtue of what are genuine disagreements possible?
On this interpretation, we must determine whether a theory of rational choice provides an answer to these questions. The questions themselves do not provide any reason to prefer one theory over another unless the theory does not answer these questions. The particular answers given cannot be grounds for preference. So, can a simple OA provide answers to each of these questions? A simple OA does identify values as relevant to rational choice, in fact simple OAs identify values as the only relevant factors in a rational choice. Thus simple OAs can respond to question (1*). Since simple OAs do not require any other content beyond the values and the weight of those values, the answers to (2*) and (3*) are that no content beyond values and their weights are needed for rational choice. Finally, question (4*) can be answered by saying that given the definition of genuine disagreement, no genuine disagreements ever arise for simple OAs. A genuine disagreement occurs when two or more people agree on the additional content that is relevant to choice yet disagree about what is the rational choice. Since on simple OAs there is no additional content, there can be no genuine disagreements. On the feature-as-question-identifier interpretation, then, even simple OAs count as prima facie acceptable theories of rational choice because simple OAs have theoretically motivated responses to each question.
The CVA also would count as prima facie acceptable because it can also provide responses to questions (2*) – (4*). According to CVA, a comprehensive value provides content beyond circumstances, particular weightings of values, and explains the existence of genuine disagreements. Thus, Chang can claim that CVA is prima facie acceptable. The features, when interpreted in this way, can do no more than indicate that both CVA and a simple OA are prima facie acceptable. Specifically, on this interpretation, the features give us no reason to prefer CVA to a simple OA. The problem is, that the features, so understood, are so weak as to allow almost any theory of rational choice in, even theories that seem to barely count as theories. Consider the “Nike theory” of rational choice that boils down to the idea “Just do it” where “it” can be any choice whatsoever. On this theory, the only bearer of positive value is the act of making a choice and the only bearer of negative value is not choosing. No other content is needed to determine a rational choice, and no other weight beside the absolute weight attached to choosing simpliciter is relevant to ascribing to a choice the honorific “rational.” There can be no genuine disagreement since there is no content that is relevant besides making choices. So, even the Nike theory of rational choice would count as prima facie plausible on the feature-as-question-identifier interpretation. Such a result is problematic because it shows that this interpretation is too permissive. It is hard to see how, on this interpretation, any theory would not count as prima facie plausible. But if it lets all theories in, then the argument does not generate any interesting results. This does nothing to help establish CVA as a viable alternative to OAs.
There are at least two ways of interpreting the features identified by Chang. On the feature- as-criterion interpretation, Chang’s arguments for features (2) – (4) do not provide the necessary independent reasons for accepting the features. On the feature-as-question-identifier interpretation, the features are so permissive as to allow almost any idea put forward as a theory of rational choice prima facie acceptable. If we are to have any reason whatsoever to view CVA as preferable to the Nike theory of rational choice, we need a different set of arguments for relying on the features is either unjustified or unhelpful. Another key argument that Chang puts forward relies on the problem of fragmentation. Unfortunately, this argument also has problems.
IV. The Problem of Fragmentation
The problem of fragmentation concerns whether we can provide a justification to believe that some conflicts are resolved by comprehensive values while others are not. Defenders of the OA may provide a fragmentary account by denying that comprehensive values are always involved in conflict resolution and rational choice. Chang challenges such views to provide “an explanation of why some conflicts are handled by a more comprehensive value while others supposedly are not. In the absence of an explanation, we have a good prima facie case for the nameless value [i.e., comprehensive value] approach.” 52 Thus, the problem of fragmentation comes down to the ability to explain in some non-ad-hoc way why some conflicts appear to be resolved by comprehensive values while others do not.
The bite of this challenge is that Chang’s version of CVA avoids this problem by holding that every choice situation involves a comprehensive value that will resolve the conflict between two other values. In “All things Considered” she states her position as follows:
My suggestion is that even when the things considered appear to be very different—cost, taste, and heathfulness; utility and maximum; simplicity and explanatory power—an all- things-considered judgment that gives each of these considerations it proper due does so in virtue of a more comprehensive value that has the things considered as its parts. 53
Chang’s CVA has plausibility because it can provide an answer as to why some cases of conflict are resolved by comprehensive values—all choice situations are resolved by comprehensive values. She also has an answer to why it sometimes appears as if some conflicts are not resolved by comprehensive values—it is an illusion that there exist conflicts that are not resolved by comprehensive values. This illusion is created because many comprehensive values are nameless. Her CVA avoids the problem of fragmentation because it does not actually countenance a fragmented account of rational choice. By avoiding this problem, Chang’s CVA should be seen as a prima facie plausible theory of rational choice.
There are three concerns about this attempt to establish the prima facie plausibility of Chang’s CVA. First, there is reason to believe that Chang’s solution to the problem of fragmentation is unsatisfactory because in setting up the problem of fragmentation she relies on an unwarranted assumptions. Second, Chang fails to demonstrate that comprehensive values are actually more than composite values. These conclusions suggest that Chang has failed to support her cautious conclusion for the plausibility of CVA. A third problem is that the move she makes to help CVA avoid the problem of fragmentation has a sibling argument that defenders of OAs can use avoid the problem of fragmentation, thus the argument cannot be used to support her bold conclusion.
Before critically analyzing the success of Chang’s argument, it is important to further clarify the distinction between the problem of fragmentation from the problem of the unity of value. 54 The key to recognizing the distinction between the problem of fragmentation and the problem of the unity of value is that these two issues are focusing on different questions. The question at the heart of the problem of fragmentation is how does a theory approach conflict resolution in general—does the theory provide a single method or multiple methods; the question at the heart of the problem of the unity of value is what explains the normative relations of the values of a particular judgment. Suppose that you have conflicts of values in two choice situations. In one choice situation, the conflict is resolved with a comprehensive value, in the other no comprehensive value, not even a nameless one, is at work. In the first situation, the problem of the unity of value is addressed because a comprehensive value provides an explanation of the normative relations between values. In the second situation, provided that one can explain the normative relations without reference to comprehensive value, the problem of the unity of value is also addressed. However, the theory that would have this result would be fragmented because it would allow for conflicts to be resolved in a multiplicity of ways. Conceptually, then, the problems of fragmentation and the unity of value are distinct, even if Chang turns out to be correct that there can be no explanations of all-things- considered judgments without recourse to a comprehensive value. 55
What are the unwarranted assumptions made by Chang in setting up the problem of fragmentation? In short, for the problem to even exist, she must establish both that comprehensive values exist and also that the moral and practical realms, individually and jointly, are not fragmented. Chang does not give us a direct reason to believe that comprehensive values exist, but she does rebut several attempts to show they don’t exist at the end of “Putting Together Morality and Well- Being.” 56 Her arguments here are sound. The fact that we don’t have names for comprehensive values does not show that they do not exist; the fact that we have difficulty explaining the content of comprehensive values does not show that they do not exist. However to assume that comprehensive values do exist based on these arguments is to commit the fallacy of ignorance. Thus without a positive argument for the existence of comprehensive values, there is no reason to accept their existence and without their existence, the problem of fragmentation may be a pseudo-problem.
One might think that Chang’s comments about Frankenstein values could get her some traction in demonstrating the existence of comprehensive values. Chang describes the concern about Frankenstein values and offers a rejoinder in the following way: opponents of CVA could be concerned that “once we allow nameless values, the floodgates are open for values to be put together in any old way. But this worry overlooks the fact that values cannot be stipulated weightings of any values whatever.” 57 In other words, a Frankenstein value is a value that contains the relevant values of a choice situation where the weights of the relevant normative relations between the values are stipulated. A comprehensive value does not have stipulated weights for values. Thus comprehensive values are not Frankenstein values. This may deflect a concern about the nature of comprehensive values, but it cannot establish that they exist because Chang must also distinguish her comprehensive value from a composite value. A composite value is a value that is an agglomeration of the relevant values without adding any new unifying content such as stipulated weight to normative relations. In this sense, a composite value is not a real entity, but a placeholder for its various contributing values. Sophisticated OAs can adopt composite values and claim that the principles or purposes provide the necessary unifying content to a judgment. Thus, the normative relations are determined by the values combined with principles. If a comprehensive value is nothing more than a composite value, then comprehensive values do not exist as entities above and beyond their components. Thus, if Chang wants to use the problem of fragmentation to establish the prima facie plausibility of CVA, she needs to explain that comprehensive values are not mere composite values because the plausibility of her view requires that comprehensive values do more than act as a placeholder. So far, she has not made arguments establishing this distinction, and thus she has not made a case for the prima facie plausibility of her CVA because she has not done enough to establish that comprehensive values exist.
Even if Chang is granted the existence of comprehensive values, she has failed to establish that the moral and the practical realms are unified. She must also do this to demonstrate that the problem of fragmentation is a real problem. Suppose that comprehensive values exist but there are only a few of them, and they do not cover all conflicts. In many other cases there are conflicts that are nevertheless resolvable. If these conditions hold, then we would need a fractured account of rational choice and conflict resolution, not a non-fractured account. If this is true, then avoiding the problem of fragmentation is neither desirable for an acceptable theory nor a realizable goal.
Notice, however, to claim either that moral reality is such that a theory should be non- fragmented, or that moral reality is such that a theory should be fragmented, is unwarranted without argument or other evidence. Similar things can be said about the practical realm. What has Chang to offer us in terms of reasons to accept the claim? Unfortunately, very little. There is no argument in either paper that attempts to provide independent reasons to believe that the moral and practical realms are such that a conflict between two values favors a non-fragmented account. The closest she comes to such an argument is another hiring example. She begins this argument with the following claim: “Start with the thought that many value conflicts have a straightforward rational resolution, and, in many of these, it is perfectly clear that the resolution is determined by a more comprehensive value that gives what matters in the choice.” 58 Two persons Aye and Bea are final candidates for a position in a Philosophy department. Aye is “original but a historical troglodyte” and Bea “is singularly unoriginal but is a bit more historically sensitive.” 59 The rational choice is Aye over Bea because originality outweighs historical sensitivity and “these weights are determined by a more comprehensive value, namely, philosophical talent, which gives what matters in the choice and determines the normative relations among its component values.” 60 Notice, this is not an argument that the case was in fact resolved by a comprehensive value. Chang merely asserts that a comprehensive value called “philosophical talent” is involved. This assertion can merely be denied as a poor explanation of this case even if one grants that comprehensive values exist. If we take the problem of fragmentation seriously, we need a way to identify which choice situations are governed by comprehensive values and which are not. Chang’s move is to assert that the comprehensive value solution is “natural and intuitive.” 61 However, this is a reliance on intuition that should be avoided because it provides no way of adjudicating between different intuitions. It is plausible that someone does not see a comprehensive value involved in this choice between job candidates. Thus this example fails to support Chang’s conclusion that the problem of fragmentation is any more than a pseudo-problem.
For the sake of argument, however, let us suppose, contra all the above arguments, that avoiding a fractured account of conflict resolution is a desirable of a theory of rational choice. 62 Can the problem of fragmentation be used to support the bold conclusion that the OA is unacceptable? Chang believes that to overcome concerns about offering a fractured account of rational choice “the orthodox approach owes us an explanation of why some conflicts are handled by a more comprehensive value while others supposedly are not.” 63 This challenge is unfair to the defenders of OAs. An OA does not have to explain why some but not all conflicts are handled by a more comprehensive value; the defender of the OA can avoid the problem of fragmentation by arguing that no conflicts are handled by comprehensive values. 64 This is the sibling move to Chang’s claim that all conflicts are resolved by a comprehensive value. A defender of a sophisticated OA would probably be better off denying Chang the first move in this argument, namely the claim that some conflicts are resolved by a comprehensive value. A defender of a sophisticated OA still has to provide an explanation as to how conflict can be resolved without reference to a comprehensive value.
To see how a defender of the OA might explain this let us consider again the example of whether Aye or Bea should be hired for a faculty position. Aye is an original thinker but weak on history, while, Bea is unoriginal but more historically sensitive. Chang believes that Aye is the better candidate. A defender of the OA can agree with all of this and provide an explanation without recourse to comprehensive values. To do this, the defender of the OA could rely on the distinction between external and internal circumstances. An external circumstance determines “which choice situation one is in, including which values are at stake and which circumstances are internal to the choice situation.” 65 An exclusionary principle is at least one type of principle operating on the external circumstances. Exclusionary principles indicate how one factor eliminates values from bearing on the choice situation, thereby eliminating the justificatory relevance of these other values. For example, consider the choice about what wine to bring to a dinner party. Let us suppose that you know the host is serving steak. An exclusionary principle might state that no white wine should be considered—it is generally assumed that only red wines should be served with steak. Thus, the factor of a steak dinner excludes the bringing of white wine. In the hiring case there could be an exclusionary principle which states “when considering job candidates for the philosophy department, historical sensitivity is irrelevant when philosophical originality is considered relevant.” This would explain why only one value, namely philosophical originality, seems to make the decision, and in this case there is no need for some new comprehensive value to be created—the rational choice is determined by the fact that the only relevant value is originality as identified by the exclusionary principle.
A defender of a sophisticated OA does not have to rely on exclusionary principles only, since there are times when exclusionary principles may not be sufficient to indicate that only one value is relevant. In these cases, one may have to look toward internal circumstance principles that “play a role in determining the relative weights of the values at stake once the choice situation has been identified.” 66 Consider again the situation in which you must bring wine to a steak dinner. The exclusionary principle only eliminates white wines, but it does not tell us anything about what red wine would be more appropriate. Do you bring a Hungarian Egri Bikavér which is blended table wine, or an Argentinean Malbec, which is a varietal from a grape that originated in Bourdeaux? If you have an internal circumstance overriding principle that says “when choosing wines, any wine that is neither a blend nor table wine is preferable to a wine that is both,” the clear choice would be to bring the Malbec. The key here is that an overriding principle in the internal circumstances does not say that the quality of wine is not to be considered at all in making a choice, just that in such a case, the nature of the wine independent of the quality will override some choices. Discount principles are specific types of overriding principles so in the case of Aye and Bea, we might have the following internal circumstance discount principle: “when considering job candidates for the philosophy department, historical sensitivity is to be discounted when philosophical originality is considered relevant.” If we have exact weights assigned to each property and we have a specific discount rate, this principle can explain the choice of Aye over Bea without making any reference to a comprehensive value. This is a plausible explanation of the situation. Similar alternative explanations could apply to all choice situations in which Chang believes a comprehensive value is at work. If this is correct, then the defender of the OA can resolve the problem of fragmentation by pointing out that she does not offer a fragmentary account of rational choice—values and principles are sufficient to generate rational choices. This claim is at least as plausible as saying that all choice situations are determined by a comprehensive value. The result of such an explanation is that the problem of fragmentation cannot be used to support Chang’s bold conclusion.
V. The Problem of the Unity of Value
The final issue discussed by Chang is the problem of the unity of value. If comprehensive values can provide an interesting solution to this problem, that solution can be used to support her cautious and bold conclusions. Comprehensive values are defined more or less as the solution to the problem of the unity of value.
More comprehensive values have a “unity” in virtue of which normative relations among their component values can be determined in different circumstances. This unity is that in virtue of which, say, a particular cost counts more than a particular moral duty in one set of circumstances, while the very same cost counts less—or less than it did, in another set of circumstances. 67
In other words, comprehensive values put the relevant values together in a way that creates a unity and this unity is achieved without stipulating weights or relying on principles. Unfortunately, this definition of comprehensive values only tells us what they do and does not explain how this unity is achieved. If Chang can explain how this unity is achieved, then she can establish her CVA as a prima facie plausible account of rational choice. Unfortunately, Chang does not give us a direct explanation about how comprehensive values create unity, but a metaphorical one about jigsaw puzzles. This final section examines this metaphor.
Since a comprehensive value is to provide the unity of a particular judgment without relying on mere weightings of values, Chang needs some way of expressing this unity. She uses the metaphor of a jigsaw puzzle to illustrate the difference between a comprehensive value and a mere weighting of values. 68 According to Chang there are at least two types of jigsaw puzzles: (a) those whose “pieces then fit together simply by their shapes interlocking in the right way” and (b) “those that “are put together in virtue of a unifying picture.” 69 She points out that the mere weighting view is similar to jigsaw puzzle type (a). Her view, however, is like jigsaw puzzle type (b). The picture “guides placement of its component parts, and it is in virtue of that picture that its parts are normatively related as they are.” 70
We can interpret this metaphor in a strong or weak fashion. The strong interpretation makes comprehensive values identical to the picture painted on the puzzle pieces—as the picture determines the spatial relations between the pieces, the comprehensive value determines the normative relations between the values. The problem with this strong interpretation is that it entails that Chang’s view violates feature (4) of theories of rational choice on the feature-as-criterion interpretation because this strong interpretation of the metaphor leaves no room for genuine disagreement. If we agree that the picture on the box is an indicator of the proper way to put the puzzle together, there can be no disagreement about what the final picture should look like—it better look just like the picture on the box. For example, if, when you put the puzzle together, the windows have door handles and doors have light bulbs stuck in them, then there can be no disagreement about the fact that you put it together wrong when the provided solution of the finished puzzle has door handles on doors and light bulbs in light fixtures. The sample solution, and thus the picture, so wholly identifies how the puzzle fits together that there can be no genuine disagreement about the correct way to put the puzzle together. If we interpret the metaphor in this strong fashion, then her view cannot account for why “there can be genuine disagreement about what the correct weighting of values is,” because the picture, as a metaphor for the comprehensive value, does not allow for such disagreement. 71 Since the possibility of genuine disagreement is a feature of choice situations that she mentions, the strong interpretation of the puzzle metaphor would indicate that her view runs afoul of this feature.
Figuring out how my strong interpretation causes this problem illuminates an underlying thrust to Chang’s criticism of sophisticated OAs. The metaphor, taken in this strong sense, describes comprehensive values such that they have a property in common with the objectionable versions of sophisticated OAs. She states that the theory must account for content beyond a particular weighting of values. Why is this? She states that the values “at stake and the circumstances in which they figure underdetermine the way in which those circumstances may affect the relative weights of those values.” 72 Now, resolving the problem of underdetermination is certainly one reason why we would need extra content. But another reason is to avoid precluding genuine disagreement. The idea is that we need to find a way that eliminates underdetermination and yet leaves room for disagreements after principles have been agreed upon. Think about how Chang might respond to my example of what wine to pair with csirke paprikás and how a person who defends a sophisticated OA can allow for genuine disagreement. She might claim that the only reason the principle allows for genuine disagreement is because the principle as stated lacks precision that can be remedied. Once this lack of precision is corrected, the defender of a sophisticated OA is on the horns of a dilemma—either he has to admit that application of principles is underdetermined, and thus a sophisticated OA needs more than principles, or the defender of the sophisticated OA has to admit that the principles resolve the conflict in only one way, and thus the view precludes genuine disagreement. Why does this preclude genuine disagreement? In short, the normative relations amongst the various values are given to us by the principles. The problem is that the resolution of conflict is somehow external to a person who is trying to make a rational choice. But, on the strong interpretation of the jigsaw puzzle metaphor, Chang’s view has the same problem: since there is a given fact of the matter about the spatial relations of the pieces—namely, the picture—genuine disagreement is precluded. If the comprehensive value is identical to the picture, then the comprehensive value is in some sense given to us in the choice situation. But notice that it is the fact that the picture or the unifying aspect of the choice situation is given to us that precludes a disagreement. And that means that Chang’s CVA also violates feature (4) on the feature-as-criterion interpretation.
Chang may claim that I have changed the definition of “genuine disagreement” in this example, and she would be correct. I am implicitly defining a genuine disagreement as one in which you agree upon the unifying component of choice situation and yet still have a disagreement about how to resolve a conflict. This is not the same as defining a genuine disagreement as one in which we agree upon principles yet still have a disagreement on how to resolve a conflict. However, Chang has gerrymandered a technicality into the definition of genuine disagreement to move us into accepting the need for something more than values, circumstances and principles to resolve conflict. A defender of a sophisticated OA would hold that the principles provide the final unifying component to the choice situation. If this is correct, then why be specific and say that it is agreement on a principle that is necessary for a genuine disagreement? The answer is that by specifying the definition this way, it requires us to look for some other way to provide unity to the choice situation. This something else would be a comprehensive value. The problem here is that she has defined genuine disagreement in a way that specifies as unacceptable a rival approach to the problem of the unity of values. This is not fair. To make the issue of genuine disagreement fairer it should be an issue that everyone confronts equally. To do this, the definition needs to be changed so that genuine disagreement occurs when the unifying component of a choice situation is agreed upon. Then we can look at the alternative responses and determine, with arguments and evidence, which solution is better. But, if this argument is correct and we modify the definition of genuine disagreement, then Chang’s CVA needs to give a response about genuine disagreements that is more than “the solution is a comprehensive value.” If the answer is that comprehensive values are like the strong interpretation of the puzzle metaphor, then her CVA violates feature (4) as a criterion. An alternative would then be to drop feature (4), but that would mean that sophisticated OAs are not eliminated as a theory of rational choice. In summary, the strong interpretation of the metaphor, along with the feature-as-criterion-interpretation, and a more fair definition of “genuine dilemma” either eliminates both sophisticated OAs and Chang’s CVA as prima facie acceptable, or does nothing to choose between them.
What about interpreting the metaphor in a weaker fashion? As it turns out, this interpretation also would preclude the possibility that there can be genuine disagreement on Chang’s CVA. On this interpretation, the picture does not identify a single solution to the puzzle but the picture also fails to provide a unifying component. Suppose that the puzzle pieces come in a small variety of shapes and that there are subtle but discernable shadings of a single color on each piece. This results in multiple acceptable ways to put the puzzle together. More than one solution is acceptable but some are clearly not acceptable. For those who might not be able to discern the subtle shadings, suppose that every piece has a number on the back so that we can check to see if two people put the puzzle together in the exact same way. For this puzzle, it would seem that genuine disagreement is possible because with multiple sensible arrangements, we could disagree about what is the solution to the puzzle. But it is hard to see how putting the puzzle together in two different, but acceptable ways, maps onto a genuine disagreement. This can be seen by considering Buridan like cases of choosing between two different piles of hay that are of the same size and quality. 73 According to Onora O’Neill, one does not choose between these piles, but merely picks one. A mere picking “is no more than a matter of lighting on some act that fits the norm under consideration.” 74 This is different from identifying the rational solution to a choice situation where some sort of rational analysis will uncover meaningful differences between options. In the Buridan case, since there is no meaningful difference between the two piles of hay, there can be no rational analysis explaining the picking of one over the other. I may pick the pile of hay on the left and you the one on the right, but there cannot be any genuine disagreement about such pickings because they are pickings and not choices. Similarly, I may solve the puzzle in one acceptable way and you in another. We can tell that they are different by turning the puzzles over and seeing that, based on the numerals assigned to pieces, that the pieces are arranged differently. But, by assumption each are acceptable solutions. I merely picked to put the puzzle together in one way and you in another. However, such a picking is not the same as a rational choice in which one makes an evaluation that one solution was more or less acceptable than the other. It is not that there are no differences between the piles of hay or the two solutions to the puzzles, but these differences do not constitute a basis for genuine disagreement because each option is equally good according to the agreed upon unifying component. These two ways should be seen as options that have equal claim to being acceptable solutions to the puzzle and you pick, not choose, one of these solutions. So, on the weak interpretation we again seem to preclude the possibility of genuine disagreement because there is nothing to disagree about. Thus by Chang’s own arguments her view appears to be unacceptable. The weak interpretation of the puzzle metaphor leads to the idea that genuine disagreement is not possible because in cases where there is disagreement, we are picking between options, not rationally choosing.
Part of the problem with using a metaphor here is that it does not describe how a comprehensive value provides unity in a direct fashion. But we need a direct accounting of how comprehensive values work, otherwise an element is missing to establish CVA as prima facie acceptable theory of rational choice. That element is that CVA is supposed to be distinct from simple and sophisticated OAs. The problem is that it is not clear that her CVA is distinct from a sophisticated OA. In “All Things Considered” Chang describes two OAs as follows:
The “simple” answer holds that the values at stake themselves determine their own normative relations with one another. The “sophisticated” answer holds that the values themselves are not sufficient; some additional normative item—usually a principle or purpose—is needed to account for the normative relations among them. Both views deny that there need to be any more comprehensive value to account for the normative relations of values. 75
The odd thing about this definition is that both simple and sophisticated OAs are marked by their opposition to a CVA. But, of course, these orthodox approaches were not created in response to the shortcomings of CVA, they were the assumed answers to questions about rational choice before Chang developed her CVA, hence they are the orthodox approaches. So, any articulation of these views could be presented in a different way, say in opposition to each other. Simple OAs only require values; sophisticated OAs hold that “some additional normative item—usually a principle or purpose [but not limited to these]—is needed.” 76 My bracketed addition here becomes very important. Chang’s comprehensive value is clearly “some additional normative item,” which would seem to make her CVA a version of a sophisticated OA. But, she maintains that her view is not a sophisticated OA. To do this, she has to deny my bracketed addition and say that sophisticated OAs are limited by their use of principles or purposes. Since the CVA is neither of these it is not a version of a sophisticated OA. But in actuality she has to say more. She has to tell us how to distinguish between principles and comprehensive values. According to Chang, comprehensive values “have a unity in virtue of which their component values hang together in the way that they do.” 77 So, comprehensive values have other values as components and provide unity. But consider a case where you are deciding whether to work on your manuscript or to play with your daughter. In general you recognize a normative element that says when one’s duty to one’s children—which is a reason to play with your daughter—is combined with positive consequences for her—say making her happy and providing her with exercise—then the influence on justification of these two values considered together is greater than the sum of its parts. This normative element describes a synergistic relationship between two values, a duty and the good associated with consequences. The result of this synergy is that even though the raw weight of the duty and consequences would not outweigh your obligation to finish the manuscript, the combination of the duty and consequences generate an additional synergistic weight. With this bonus weight, the consequences and duty of playing with your children outweighs the obligation to finish the manuscript. 78 Is this normative element a value or a principle? Shelly Kagan describes what he calls interaction principles which specify “how various factors interact so as to determine the moral status of particular acts.” 79 His use of the term “factor” is equivalent to the use of the term “value” throughout this paper. This normative element describing a synergy between two values could legitimately be called a principle. It also seems to provide the unity, i.e., describes the normative relations that a duty and a good related to playing with my daughter have vis-à-vis each other and explains that overall relationship with my obligation to finish a manuscript. If this is correct, then this principle is a comprehensive value. But that means comprehensive values are nothing more than the principles that have been used throughout this paper. And if that is correct, then the CVA is just a version of the sophisticated OA. This would be problematic if Chang’s arguments against sophisticated OAs are more successful then the arguments contained in this paper suggest for that would mean she has provided the arguments that discredit her own view. She has still has the option of distinguishing between principles and values in some other way, but at this point it is not clear how to make the distinction. Defenders of sophisticated OAs offer the principles or supplemental normative elements as providing unity to the values. To claim that principles or other elements are inadequate without marking a clear distinction between them and comprehensive values means that she may have argued against her own view.
- Conclusion
Ruth Chang wants to defend a comparativist account of rational choice. She favors an approach that utilizes comprehensive values to determine the rational choice in a given situation. She sets the possible choices—her approach and versions of the OA—and describes three issues that such theories should address, namely features of rational choice, the problem of fragmentation, and the problem of the unity of values. Each of these issues is supposed to provide support for her CVA. However, upon examination, there are a host of problems with the arguments intended to support her CVA against simple and sophisticated OAs. Because of these problems, Chang owes us a different set of arguments. The core of these new arguments should be a direct description of comprehensive values and how they put values together that are distinct from the principles or purposes cited by defenders of sophisticated OAs. 80
Bibliography
Chang, Ruth. “All Things Considered.” Philosophical Perspectives 18 (2004): 1-22.
Chang, Ruth. Incommensurability, Incomparability, and Practical Reason. Cambridge: Harvard UP, 1997.
Chang, Ruth. Making Comparisons Count. New York: Routledge, 2002.
Chang, Ruth. “Parity, Interval Value, and Choice.” Ethics 115 (2005): 331-50.
Chang, Ruth. “The Possibility of Parity.” Ethics 112 (2002): 659-88.
Chang, Ruth. “Putting Together Morality and Well-Being.” Practical Conflicts: New Philosophical Essays. Eds. P. Baumann and M. Betzler. New York: Cambridge UP, 2004.
Dworkin, Gerald. “Theory, Practice, and Moral Reasoning.” The Oxford Handbook of Ethical Theory. Ed. David Copp. New York: Oxford UP, 2006. 624-44.
Dworkin, Ronald. “Is There Really No Right Answer in Hard Cases?” A Matter of Principle. Cambridge: Harvard UP, 1985. 119-45.
Gert, Joshua. “Value and Parity.”Ethics 114 (2004): 492-510.
Hubin, Donald C. “Hypothetical Motivation.” Nous 30.1 (1996): 31-54.
Kagan, Shelly. Normative Ethics. Boulder: Westview, 1998.
O’Neill, Onora. “Normativity and Practical Judgment.” Journal of Moral Philosophy 4 (2007): 393-405.
Postow, Betsy. Reasons for Action: Toward a Normative Theory and Meta-Level Criteria. Dordrecht, Holland: Kluwer, 1999.
Rabinowicz, W. “Value Relations.”Theoria 74 (2008): 18-49.
Wasserman, R. “Indeterminacy, Ignorance and the Possibility of Parity.” Philosophical Perspectives 18 (2004): 391-403.
- In this paper I am speaking of justificatory choice, choice between propositions that describe actions or states of affairs which identify one option as more justified than another. This should not be confused with choices that have motivational content and lead one to act in a particular way or to seek out a particular state of affairs. Thus, even if an action is described as justified, that does not entail that an agent is motivated to perform that act.
- Chang notes that her the term “value” is to be “understood in the broadest sense of that term to include rights, duties, obligations, and standards of excellence, as well as their negative counterparts such as ugliness and badness” (Ruth Chang, “All Things Considered,” Philosophical Perspectives 18 [2004]: 6). When I use the term “value” I should be understood as using it in the same way.
- Chang also mentions two other approaches, namely the Procedural Approach and the Single-Point- of-View Approach.
- Buridan type cases where two objects are identical in their properties provide a different challenge to comparativism. In these cases, there are plenty of points of comparison between two options but for each point of comparison, the associated values are identical. In these cases, it seems as if comparativism might be frustrated as well because there are no differences between the options, so a comparison will not determine choice. Onora O’Neill has recently made comments about such cases in “Normativity and Practical Judgment,” Journal of Moral Philosophy 4 (2007): 400.
- Ruth Chang, “Introduction,” Incommensurability, Incomparability, and Practical Reason (Cambridge: Harvard UP, 1997) 4; Ruth Chang, “The Possibility of Parity,” Ethics 112 (2002): 660-1.
- Chang, “The Possibility of Parity”; Ruth Chang, “Parity, Interval Value, and Choice,” Ethics 115 (2005): 331-50; Joshua Gert, “Value and Parity,” Ethics 114 (2004): 492-510; W. Rabinowicz, “Value Relations,” Theoria 74 (2008): 18-49; R. Wasserman, “Indeterminacy, Ignorance and the Possibility of Parity,” Philosophical Perspectives 18 (2004): 391-403.
- Ruth Chang, “Putting Together Morality and Well-Being,” in Practical Conflicts: New Philosophical Essays, eds. P. Baumann and M. Betzler (New York: Cambridge UP, 2004) 119.
- Chang, “Putting Together Morality and Well-Being,” 119.
- Chang, “Putting Together Morality and Well-Being,” 119.
- Chang, “All Things Considered,” 6.
- Chang, “Putting Together Morality and Well-Being,” 131-2; Chang, “All Things Considered,” 6.
- Chang, “Putting Together Morality and Well-Being,” 133; Chang, “All Things Considered,” 6.
- Chang, “Putting Together Morality and Well-Being,” 152.
- Gerald Dworkin, “Theory, Practice, and Moral Reasoning,” in The Oxford Handbook of Ethical Theory, ed. David Copp (New York: Oxford UP, 2006) 628.
- Betsy Postow, Reasons for Action: Toward a Normative Theory and Meta-Level Criteria (Dordrecht, Holland: Kluwer, 1999) 3.
- Chang, “Putting Together Morality and Well-Being,” 152.
- Chang, “Putting Together Morality and Well-Being,” 119.
- Donald C. Hubin, “Hypothetical Motivation,” Nous 30.1 (1996): 44.
- Dworkin, “Theory, Practice, and Moral Reasoning,” 631.
- Ruth Chang, Making Comparisons Count (New York: Routledge, 2002) 29.
- Chang, Making Comparisons Count, 30-33.
- Chang, “Putting Together Morality and Well-Being,” 119.
- Dworkin, “Theory, Practice, and Moral Reasoning,” 631.
- Chang does not summarize these points as clearly in her paper “All Things Considered” but each feature is mentioned there.
- Chang, “Putting Together Morality and Well-Being,” 135-6. Chang intends these features to choose between only rival versions of comparativist theories of rational choice. These are not necessarily features required to govern the choice between comparativist and non-comparativist theories.
- Chang, “Putting Together Morality and Well-Being,” 129.
- Chang, “Putting Together Morality and Well-Being,” 129.
- Chang, “Putting Together Morality and Well-Being,” 132.
- Chang, “All Things Considered,” 4.
- Chang, “Putting Together Morality and Well-Being,” 136.
- Chang, “Putting Together Morality and Well-Being,” 135.
- Chang, “Putting Together Morality and Well-Being,” 131.
- Chang, “Putting Together Morality and Well-Being,” 131.
- Chang, “Putting Together Morality and Well-Being,” 129.
- Chang, “Putting Together Morality and Well-Being,” 142.
- Chang, “Putting Together Morality and Well-Being,” 132.
- Chang, “Putting Together Morality and Well-Being,” 133-4.
- Chang, “Putting Together Morality and Well-Being,” 134.
- Chang, “Putting Together Morality and Well-Being,” 134.
- Chang, “Putting Together Morality and Well-Being,” 134.
- Chang, “Putting Together Morality and Well-Being,” 136.
- Chang, “Putting Together Morality and Well-Being,” 135.
- Chang, “All Things Considered,” 14.
- Chang, “Putting Together Morality and Well-Being,” 134.
- Chang, “Putting Together Morality and Well-Being,” 134-5.
- Chang, “Putting Together Morality and Well-Being,” 135.
- Chang, “Putting Together Morality and Well-Being,” 132.
- Note that this does not call into question the assertion that once we have agreed upon a correct weighting of values there can be no genuine disagreement: if everyone does in fact agree upon the weights, then it is only through an error of calculation that disagreement occurs and, by definition, this is not a genuine disagreement.
- Chang, “Putting Together Morality and Well-Being,” 135.
- Chang, “All Things Considered,” 14.
- Chang, “All Things Considered,” 14.
- Chang, “Putting Together Morality and Well-Being,” 128. There is the possibility for confusion in Chang’s writing concerning the problem of fragmentation. The problem of fragmentation has two versions. On one version, when considering pure moral conflict cases, if you believe that some cases are resolved by comprehensive factors, and others are not, then you have intra-point-of-view fragmentation. The same fragmentation problem appears if you think that in pure prudential cases some cases are resolved by comprehensive factors and others are not. However, on a second version, a fragmented account is such that all pure prudential cases are resolved by a comprehensive factor, but no moral cases of conflict are so resolved (or vice versa). This is inter-point-of-view fragmentation. To this second version one might say that prudential cases can always be adjudicated in terms of utility while moral cases, because there are virtues, rights, rules, consequences, obligations, etc., preclude the possibility of there being comprehensive factors.
- Chang, “All Things Considered,” 3.
- There is the potential to conflate the problem of fragmentation with the problem of the unity of value. Chang sometimes says things similar to the following when discussing the problem of fragmentation: “The nameless value approach, in contrast, provides the basis for a unitary account of conflict resolution” (Chang, “Putting Together” 128; emphasis added). Her choice of the word “unitary” in this example, as well as other places could be the source of such a conflation.
- The problem of the unity of value will be discussed again when an analysis is put forward of the jigsaw puzzle metaphor.
- Chang, “Putting Together Morality and Well-Being,” 144-7.
- Chang, “Putting Together Morality and Well-Being,” 144.
- Chang, “Putting Together Morality and Well-Being,” 125.
- Chang, “Putting Together Morality and Well-Being,” 125.
- Chang, “Putting Together Morality and Well-Being,” 126.
- Chang, “Putting Together Morality and Well-Being,” 126.
- This supposition does not entail that we grant Chang that either comprehensive values exist or that the moral and practical realms are not fragmented.
- Chang, “Putting Together Morality and Well-Being,” 128.
- The assumption that comprehensive values do in fact exist is dropped in the argument from this point forward.
- Chang, “Putting Together Morality and Well-Being,” 142.
- Chang, “Putting Together Morality and Well-Being,” 132.
- Chang, “All Things Considered,” 15.
- Chang, “Putting Together Morality and Well-Being,” 138-9; Chang, “All Things Considered,” 17.
- Chang, “Putting Together Morality and Well-Being,” 139.
- Chang, “Putting Together Morality and Well-Being,” 139.
- Chang, “Putting Together Morality and Well-Being,” 135-6.
- Chang, “Putting Together Morality and Well-Being,” 133-4.
- O’Neill 401.
- O’Neill 400.
- Chang, “All Things Considered,” 6-7.
- Chang, “All Things Considered,” 7.
- Chang, “Putting Together Morality and Well-Being,” 136.
- As John Nolt pointed out to me, synergistic interaction principles may, on occasion, frustrate choice. A synergistic interaction principle could increase the influence of otherwise weak factors or decrease the influence of strong factors to a level that results in a tie where there was not one before. This would produce a form of dilemma instead of resolving conflict. However, the probability of this happening is quite low in comparison to a synergistic interaction principle either raising or lowering the influence of a factor away from a tie: in any given situation, there is only one point in which there can be a tie and a large number of points in which no tie occurs.
- Shelly Kagan, Normative Ethics (Boulder: Westview, 1998) 183.
- There are many people whose help made this paper better. First and foremost is my mentor, Betsy Postow. Others include John Nolt, Clancy Martin, Joseph Moore, David Concepción, Judith A. Okapal, Stephen G. Morris, anonymous reviewers, and audiences at the University of Missouri, Kansas City, the University of Nevada, Las Vegas, and the American Philosophical Association, Central Division.