Weak supervenience 6 26 02 6/26/2002 4:16 PM C:\B\B\n\Weak supervenience 6 26 02.doc
Matjaž Potrč, University of Ljubljana, Slovenia, firstname.lastname@example.org
The difficulty with the weak supervenience is in that it allows for strange behavior of evaluative terms with only a slight change in subvenient basis. This difficulty is the result of generalist presuppositions entrenched in the way supervenience is supposed to work. On the basis of these presuppositions it is claimed that strong supervenience should be followed. If one embraces the opposed particularist presuppositions though, then the importance of the weak supervenience may be revived, and it turns out to be just what is needed in an account of evaluative properties’ behavior. Weak supervenience is also compatible with the naturalistic program involving evaluative terms.
It is held that weak supervenience cannot be adequate, for it allows for strange behavior: weak supervenience holds just for the actual world in which this descriptive subvenient property basis determines supervenient properties. But if there would be just a slight difference in the subvenient basis of a possible twin world, say just one atom missing, it may be that all the supervenient properties, such as that of being good, could completely reverse their valence. What was morally good would become morally bad. But this seems inadequate. So strong supervenience is needed, which extends over all possible worlds. If certain descriptive subvenient basis results in an evaluative fact of a certain valence, then this specific valence of evaluative facts should be extended over all possible worlds.
This argument is questioned. The first questionable position is Kim’s construal of subvenient basis to consist of thick properties, which already come evaluatively laden. Even if these subvenient properties would ground the supervenient set of properties, there would be further hurdles. It is an unsupported presupposition that supervenient properties get projected over all possible worlds, as the strong or global supervenience thesis requires.
Even more, from a particularist point of view, the weakness of the weak particularism that seems disastrous from the generalist point of view (to which Kim seems wedded) is actually an advantage, going along with the particularist statement about possible change of value in the even slightly changed descriptive circumstances.
So weak supervenience is revived – as an adequate basis of particularism. A problem though is whether weak supervenience is suitable for naturalistic approach, given that non-naturalist interpretations are sometimes embraced as being compatible with particularism. It is argued that weak supervenience may be compatible with naturalistic particularism, from this perspective, and not just with the non-naturalist particularism.
Weak supervenience claims that higher order terms, such as evaluative properties, are in relation of dependency in respect to their subvenient basis. If St. Francis is good, then his higher order property of being good depends on the lower order properties, in the sense that, if there would be the same lower order property arrangement in another possible situation, then the St. Francis duplicate over there would be good as well.
Although dependency between the lower subvenient and higher supervenient orders of properties is claimed by the weak supervenience, there is no necessity of this relation affirmed. Necessity though is achieved by strong supervenience, which claims that the identity of subvenient bases has to hold along all possible worlds.
Particularism may come in several forms and one of these forms claims that moral principles can always be overridden. Particularism is a view opposed to the generalism which affirms normative authority of general principles.
It may be said that for generalism, there is a general pattern of pairings involving a list of cases involving descriptive and evaluative properties. Such a list does not need to be united by general principles according to particularism.
We may think about the extension of several cases such as
D1 → D2, …, Dn → En
to range over all possible worlds (D is for descriptive here, and E is for evaluative). In this case we would have a general pattern ranging over all these possibilities, and so we would have generalism. If contrary to this we do not think that there is a necessary transition from one case to the next, then we do not have an overall pattern tying them. Whereas the first case of systematizing seems to feature strong supervenience, the second particularist case features weak supervenience.
Strong and weak supervenience are not just neutral descriptions. Rather, strong supervenience goes along with generalism, whereas weak supervenience has a tendency of naturally fitting particularism. The reason is that the extension over possible worlds assures general pattern in the case of strong supervenience. Whereas weak supervenience only gives naturalistically respectable conditions for each particular case, without that any such general pattern would be involved.
Kim defines weak supervenience as follows:
“A weakly supervenes on B if and only if necessarily for any x and y if x and y share all properties in B, then x and y share all properties in A – that is, indiscernibility with respect to B entails indiscernibility with respect to A.” (58)
Here, B is the subvenient base, and A is for the realm of the supervenient. According to Kim, the supervenient relation of dependency is really a “relation between families of properties” (55) of which respectively each the base and realm of the supervenient consists.
Here is one example. If x and y both share subvenient properties of being courageous and benevolent, say, they are also bound to share the property of being good.
On the face of it, this seems to be a too strong requirement, for sharing all the properties means being an exact physical duplicate in the same world. And each physical duplicate has an enormous amount of properties. Besides to the ones mentioned, there are also properties of having a certain weight, a certain age, and many more. But not all of these are required if we focus our attention at the subvenient basis whose resultance is the supervenient predicate of being good. The talk about the “relation between families of properties” may surprise at first sight. But here is its explanation: the supervenience relation determines several or many properties, not only in the subvenient basis but also in the area of the supervenient. Many or most of these properties belonging to the realm of the supervenient will not be relevant for the supervenient property we center upon, being good in our case. So, the relation of weak supervenience is even too strong a requirement as supervenient resultance from the subvenient basis of properties. It is too strong, because relevance is not taken into account. But in the practical rendering of the example, which includes properties of being courageous and benevolent, just these relevant properties are given as supporting supervenient property of being good* (G*), i.e. being good in the sense of having a positive or negative valence (G, -G). Thus there is not a whole family of properties given, although this would be the requirement of weak supervenience.
“Weak supervenience of A on B therefore comes to this: any two objects with the same B-maximal property must have the same properties in A – they are both G or both –G.” (59)
Thus the weak supervenience, although it goes under the name of weak, on the one hand gives too strong requirements, and one reason is that it does not cash in the relevance. Relevance though is respected in the above formulation, as a preamble to the practical examples that illustrate it. Although there is no “family of properties” anymore that we face at the supervenient level.
But this is not the worry that Kim would address in the case of the weak supervenience. To the contrary, he claims that weak supervenience is “too weak” (59). In what sense?
What valence the subvenient basis will produce for the supervenient property, “whether G* is G or –G depends on the particular world under consideration, and is not a feature invariant across possible worlds” (60).
The requirement is thus that indiscernibility is not just limited to the actual world, but that it extends to other possible worlds, and thus that it turns out to be a necessary relation, actually that it becomes strong supervenience. Strong supervenience thus brings the necessary lawful generalization along with it.
If one would just stay with the relation of weak supervenience, the following strange things would be permitted, according to Kim:
“(a) In this world anyone who is courageous, benevolent, and honest is a good man, but in another possible world no such man is good; in fact, every such man is evil in this other world.
(b) Again, in this world anyone who has courage, benevolence and honesty is good; in another world exactly like this one in respect of the distribution of these virtues, no man is good.
(c) In another possible world just like this one in respect of who has, or lacks, these traits of character, every man is good.” (60)
Weak supervenience stays “within” (60) possible worlds, which we may interpret as possible situations, but does not extend to other global possible situations. It does not assure any pattern. There is no projectibility coming along with the weak supervenience:
“The particular association between A-properties and B-properties in a given world cannot be counted on to carry over into other worlds” (60).
Here is the reconstructed argument (compare p. 60):
Pr1 Weak supervenience only requires that within any possible world there are no two things agreeing in B but diverging in A.
Pr2 It is possible according to the weak supervenience that things agreeing in B in this world diverge in A in other possible worlds.
C1 .: According to the weak supervenience, the particular association between A-properties and B-properties in this world will not transfer to other worlds. (There is no general pattern according to the weak supervenience.)
C2 .: According to the weak supervenience, B-properties do not fix its A-properties (its supervenient properties).
C3.: According to the weak supervenience the base properties do not determine supervenient properties.
The requirements in the above argument may be elucidates with the help of particularism. Particularism claims that there is no general pattern that would extend over cases that possess a certain evaluative valence, such as being good.
From this point of view, weak supervenience thesis assures a subvenient descriptive basis for the evaluative, without that it would also assure a projectible subvenient descriptive basis for the evaluative.
Each evaluative supervenient A-property is supported by a subvenient B-property descriptive basis, and this suffices for the requirements of naturalism. But if the situation or possible world changes, even slightly, there will not be any automatic extension of the same relevant B-basis to result in the same A-properties. Similarity between situations or between possible worlds does not suffice to have an automatic transmission from one to the next situation figuring the same value of the evaluative A-property on the basis of the same subvenient B-properties. And this is just what particularism claims: that there is no pattern assured from the descriptive to the evaluative.
The strange consequences (a), (b), (c) are exactly what the particularist agrees with. Possessing a set of properties such as being courageous, benevolent and honest may change their valence in another situation (in another possible world). The reason may be that the relevance (which is not thematized by Kim for the case of weak supervenience) does not extend just to the three enumerated subvenient B-properties that result in supervenient A-property (good, say). The relevance actually extends to the whole of each new situation, of each new world. And it is then not so surprising at all that the valence may be changed in the new situation. Similarity is not enough to assure the sameness of the valence of supervenient A-property. But there is also a strange requirement here if one strengthens the situation from one case to the next: then it seems that the possible worlds would actually need to be identical in their entire overall subvenient B-properties base. But this seems to be a too strong requirement even for strong supervenience, which should after all leave the space for difference between different worlds or situations. At least something in situations (worlds) should be different so that they would differ from each other.
Pr1 presents a minimal naturalist requirement. If there are exactly identical B-property bases in a situation (within a world), then there will be no divergence in supervenient A-properties. This minimal naturalist requirement though does not take into account the relevance – which properties are exactly these properties in a situation such that they determine the supervenient A-property. Determination relation may be holistic.
Pr2 seems to present a realist statement, at least from the point of view of the particularist. Say that there is a situation (a world) in which a small, finite or tractable number of B-properties diverge in the outcome about which A-property gets produced. If there is one situation where there are C, V and H in the subvenient basis, and they produce good as their supervenient outcome in this situation, the supervenient outcome may be just the opposite of good in another situation (world) figuring C, V and H. The relevance must in this case of course be extended over all the holistic possible situations. With this, we get a classical setting of the particularist. Even if the kernel stays with the same properties in the subvenient base, the holistic environment of the situation may still change and so it may contribute to the result bringing in new valence to the supervenient property.
C1 This conclusion is denying the extension of patterns for the same subvenient basis. But we have seen that the difficulty in determining the subvenient basis was the one about relevance. As many features may be relevant, particularist bites the bullet and transfers the relevance to the whole holistic situation. Each situation will have its own evaluation. As per definition each two situations differ (otherwise they would be the same situations), the holistic system may have different impact in each of these situations. But this just means that there is no general projectible pattern available.
C2 It is not justified to conclude from the above though that “B-properties do not fix the A-properties”. Certainly they do not fix them in a rigid way. But this would be unrealistic anyhow. We will admit that the moral and other evaluative situations change, and that because of this change, there may be different evaluations in different situations, even if some limited number of B-properties gets preserved from one complex situation to another one.
C3 follows even less from the previous points in the argument than does C2. If “determining” is to be understood in the sense of “completely determining” this may be an appropriate characterization. But complete determining would seem to be too strong anyhow. It is indeed a requirement steaming from the strong supervenience. And it is dubious whether strong supervenience is flexible enough in order to assure a proper working and explanation in the area of of moral deliberation and in the area of deliberation supporting action. Determination in the sense of weak supervenience may be holistic and thereby broader, not based on atomistic presuppositions. It seems that just this kind of determination is also realistic in supporting our moral and action directed deliberation.
The overall result of the above discussion is that weak supervenience is seen as a defunct from the generalist perspective, which would require strong supervenience. We may wish to put things in order to some extent, by analyzing the situation.
1. First, what is it that the weak supervenience is committed to?
a. There is naturalistic basis: non-moral subvenient underpinning of the moral. I.e. if there would be another subvenient basis just like this one, it would result in the same kind of supervenient property.
b. The supervenience relation just holds for this world. I.e., if there would be another subvenient basis situation just like this one, it would have to exist in this world in order that the supervenient relation would hold.
2. What are the reasons weak supervenience is rejected?
a. Disagreement with (1.b). I.e., if in determining the supervenient relation one stays in this world only, the supervenient properties may show a strange behavior in other, possible worlds.
3. Questions directed at the reasons for rejection:
a. Strange behavior of weak supervenience described in (2.a) is strange on the basis of the presupposition only that strong supervenience is the correct position.
b. The difference between strong supervenience and weak supervenience is that the first one pushes in direction of generalism, i.e. it extends over all possible worlds. And extension over all possible worlds is exactly the move promoted by generalism. But this move is dubious, because possible worlds must be at least slightly different to the actual one (otherwise they are identical to it).
Whereas weak supervenience only secures subvenient basis for each particular occasion of the supervenient property appearance. But this results in a strange behavior once it is transferred to other possible worlds. Transfering conditions for subvenient support of the supervenient properties assures a pattern, i.e. generalization of these conditions. But such a pattern is denied by weak supervenience. Weak supervenience displays strange behavior, strange if measured with the generalist strategy (atomism and tractability in the subvenient basis) of strong supervenience model. The same behavior (of not extending the conditions of subvenient basis for a certain supervenient property to other possible situations or other possible worlds) is not only compatible with, it is even desired by the weak supervenience. So, weak supervenience is compatible with particularism.
The presupposition of generalism are as follows. The subvenient basis in the actual world is atomistic and of tractable nature. This should be countered by particularism: the subvenient basis in the actual world is holistically complex, nontractable and not atomistic.
The definition of weak supervenience as it actually stands is shaped by the generalist presuppositions, so that the entire score is biased in favor of generalism, and in favor of strong supervenience as the model of generalist relation.
Therefore, the very definition of weak supervenience should be shaped otherwise, so that it will be based on the rich holistic nontractable nonatomistic basis. What is needed is thus a new definition of weak supervenience as Particularly Weak Supervenience. As this Particularly Weak Supervenience will be there in order to catch the holistic nontractable nonatomistic subvenient basis and the special way this basis supports supervenient properties, without generalist presuppositions, it may be compatible with local supervenience definition (perhaps along Horgan’s local supervenience definition and the overall idea in Superdupervenience paper). And if PWS (Particularly Weak Supervenience) is related to nontractable holism, including its vicinity to Local Supervenience, then PWS is also the right basis for naturalist subvenient basis of particular quasi-items or quasi-entities as encounterd in the Blob, one complex nontractable holistic world. All this merits to be elaborated. Let us turn to the question about the relation between weak supervenience and naturalism now.
Particularism is a naturalistically supportable teaching, under the condition that it embraces weak supervenience. For weak supervenience does not promote any projectability and thereby it does not buy any generalities, which is compatible with the definition of particularism.
Embracing of non-naturalism, to the contrary, seems to succeed under the presupposition that strong supervenience and thereby generality is adequate for particularist enterprise. But this just cannot be right. So as a particularist, especially as a naturalistically minded particularist, one should reject strong supervenience with its underlying generalist presuppositions, and one should embrace weak supervenience.
Some supporters of particularism are wedded to strong supervenience and thereby they are committed to generalist presuppositions. For generalist presuppositions embrace projectability as a precondition for a viable non-naturalist project.
But there is no need for projectability (extension of a property and of its value over possible worlds) in order that a naturalist project would be pursued. So, particularism is not saved by the move in direction of non-naturalism, it is just pushed into an implausible direction. Non-naturalism seems to be an implausible direction, for there does not seem to be any explanation possible along its lines. It is also not taken into account that weak supervenience requires non-generalist explanation, in form of complex narrative stories, and that these figure as reasons, which is incomprehensible to the generalist.
Some defenders of particularism are still prey to the generalist presuppositions. They think that particularism is tied to non-naturalism. But the push towards non-naturalism is promoted by the presupposition that the right supervenience is strong supervenience, one that allows for projectability. And projectability is tied to the generalism, so by embracing the presupposition of strong supervenience the promoters of non-naturalism are really continuing to support generalism.
As against this, particularism seems to be compatible with naturalism. One option of how to advance the project of naturalistic particularism is to follow weak supervenience instead of strong supervenience.
Is there a “connection between moral and non-moral properties”? Little (2000) denies it and says that as related to this, “the model backing particularism clearly belongs in the non-naturalist camp”(4/5). But what should the connection between moral and non-moral properties really look like? Should it be covered by general patterns, patterns that extend over all possible worlds, as the strong supervenience requires? Or should there be a naturalist basis for moral properties, and just this, without extension to further cases, as this is required by weak supervenience? It seems that this last option widely exceeds the first one for anybody that would like to have both particularism and naturalism. And this combination seems to be a desirable outcome, also because naturalistic particularism offers a support for a viable explanatory story, which is precluded to the non-naturalist approach.
What is the meaning of moral properties being “shapeless with respect to the nonmoral” (5)? Does this entail that moral properties should be without any generalized pattern over instances or possible cases? Yes. Does this require that there be no non-natural basis for the supervenient moral properties? No. Notice now that extending non-moral to moral pattern over a range of possible situations or worlds is required by strong supervenience thesis, and that there are reasons for this being unjustified. Notice on the other hand that weak supervenience thesis distinguishes two tasks: (1) Providing a nonmoral naturalist basis for each instance of the moral. (2) Not extending this basis to other possible situations or possible worlds. Now, (1) is entirely adequate for a naturalist particularist project. Whereas (2) excludes strong supervenience and therewith it excludes generalism. Anyway, “being shapeless with respect to the nonmoral” does not seem to involve any requirement that there would be no naturalist base for case-to-case examples. But this is all that particularism really requires. Asking for more, i.e. for extension of cases over several situations or worlds may only come from a generalist presupposition.
Little (2000) argues in favor of non-naturalistic particularism by using two considerations. Let us observe how the presupposition of projectible and generalizable patterns, that is actually a property of generalism and a presupposition of strong supervenience, is shaping up the following reasoning.
(i) The first reasoning in direction of non-naturalistic particularism builds on the impossibility of extending to the general patterns. Here is the reconstruction of the argument:
Pr1 There is no uniform extension from each non-morally supported particular case to another such case.
C1 .: Therefore, there is no non-moral supporting of the moral cases.
C2 .: Therefore non-naturalism follows.
The argument overlooks that moral cases may still be non-morally supported, for each case, even if there is no general pattern around, as Pr1 claims. So the intermediate conclusion C1 seems to be misguided. Notice that it does not follow that there cannot be any support of the non-moral to the moral if there is not any generalizable pattern extending a case over several possible situations. From this point of view, the following seems to involve a strangely unfounded reasoning:
“One reason for advancing a doctrine of shapelessness is the belief that we cannot mark out the boundaries of moral concepts in purely nonmoral terms: the items grouped together under a moral classification such as ‘cruel’ do not form a kind recognizable as such at the natural level. The thought here, familiarly, is that, of the infinitely many ways of being cruel – kicking a dog, teasing a sensitive person, and forgetting to invite someone to a party might each qualify – there is no saying what they have in common (and why, say, the pain inflicted during a spinal tap is different) except by helping oneself to the moral concept of ‘crulety’. This is to believe non-naturalism, because one denies there are any finitely-specifiable conditionals of the form ‘If M then N’.” (5)
Having something in common, it seems, is not a requirement for non-moral support of the moral. Each case of cruelty may well be naturalistically supported, even if there is no general pattern covering all of these cases. So there is no reason to adopt non-naturalism if one is a particularist. There is just one conditional ‘If M then N’ for each particular case. No projection from one case to the others is needed to preserve both particularism and naturalism. One may ask oneself though why the requirement of this generalizable extension is posed besides to the viable case-to-case non-moral naturalist support of moral property. It is hard to imagine something else as the adoption of the presupposition that the strong supervenience gives an appropriate account of the relation between the moral and the non-moral properties. To the contrary, kicking a dog, teasing a sensitive person and forgetting to invite someone to a party are each one a case of cruelty, and they are appropriately non-naturalistically supported (each by another set of properties, still furthering the diversification between cases by taking a different holistic background into account for each case), although they do not form a pattern common to them all. The question prompting the recourse towards generalization seems to be “What makes all these particular cases examples of ‘cruelty?” But this question does not have any direct link to the support of each case by a non-moral basis. This basis may stay, although it would not be generalizable over a range of instances.
(ii) The second reasoning in favor of non-naturalistic particularism seems more decisive to Little. Particularist
“denies that we have reason to hope for any finitely-specifiable conditionals of the sort ‘If N then M’. … The point is to deny that such considerations [of each moral case being supported by the non-moral in case of particular instances] carry their reason-giving force atomistically. Natural features do not always ground the same moral import… Holism is not complicated monism… There is no cashing out in finite or helpful propositional form the context on which the moral meaning depends… This is not to deny that the moral supervenes on the nonmoral.” (6)
Thus, as it was already established in the discussion of the former point, the moral may without harm be supported by the nonmoral properties for each case. The problem is then just that each such case remains holistically bound to a specific context. The radical difficulty here about the particularism is that there cannot even in principle be identified any subveniently contributing properties, as atomistic properties, because the contribution of subvenient properties is deeply contextual for particularist, and so it is bound to be different from one case to another. There is no general pattern unifying these particular cases where a very weak form of supervenience gets established. But why would one need something more than the weak supervenience?
The problem is that for a “particularist… the conditions for something being cruel can’t be spelled out … at all” (8). They cannot be spelled out because spelling out would require a tractable procedure, based on the contribution of subvenient properties that would be determined in an atomistic way. But this is not possible if particularism is really bound to a radical holism in respect to the contributory causes.
It is hard to see again why this all would push someone into the direction of non-naturalism. Little herself admits that each case of cruelty, say, is appropriately and uncontroversially subveniently naturalistically supported. Just that there does not exist any projectible pattern that would cover all cases of cruelty in respect to their subvenient basis. So she admits naturalism as to the instances of cruelty (naturalistic holistic support for each case), but denies naturalism about a projectible subvenient basis supporting cruelty. It is not clear though why this subvenient projectible basis would be necessary for naturalism. Naturalism may well succeed for each particular case in a holistic nontractable manner. Just that this does not imply any necessity of having more than weak particularism – weak particularism allows for the first particular nonmoral underpinning of the moral already. Only strong supervenience may incite towards this stronger atomistic and tractable subvenient support requirement.
So, what kind of supervenience is adequate for particularism: weak or strong? The answer seems to be not just that weak supervenience is an alternative that particularists usually take; it is even mandatory for particularists to take weak supervenience if they wish to preserve their status of particularists. The very notion of weak supervenience needs to be altered though so that it will not be shaped anymore by the generalist views that take strong supervenience as a model. The argument that strong supervenience is a must needs to be resisted. Particularism shows that weak supervenience is appropriate for moral properties as interesting or relevant supervenient properties.
Moral Twin Earth argument claims that the behavior of the supervenient moral properties is different to the behavior of supervenient mental properties. How is this related to the fact that subvenient base for the moral is circular (courageous, benevolent), where physical basis of the mental seems not to be naturalistic indeed? This may be used in order to clarify the fact that Kim proceeds with thick already morally loaded properties (courageous) in the subvenient basis.
What is the change of supervenience base that allows for so called disastrous consequences with weak supervenience?
a. If the basis is that of thick properties (already loaded properties, such as “courageous”, “moderate”), there is a question about how such a basis may be changed at all. If there are changes here, these are already changes in the evaluatively loaded properties.
b. Another possibility is that changes are in the physical basis: one atom changes in the next possible world, in respect to the actual world, and possibly all evaluative properties go berserk by changing their valence. (There is a question about the relevance of this.)
Perhaps a. and b. are just possibilities on a large spectrum of things. Perhaps a. is not plausible because of its circularity, and b. is not plausible because of being of the physical base nature and too far away from the evaluative. How can then one establish the desired connection?
Consequences actually introduce holistic picture and this is why they seem a nuisance to Kim, who subscribes to two generalist presuppositions of i. atomism and ii. tractability.
All this is based on an utterly wrong presupposition about how the supervenience works. The subvenient basis just cannot be atomistic and tractable in its nature.
The argument against weak supervenience:
Pr1 Weak supervenience assures the dependency of supervenient on the subvenient basis.
Pr2 Yet because the weak supervenience is limited to the actual world only, small differences in subvenient basis may have extensive consequences for value changes in the supervenient evaluative area in the next possible worlds.
Pr3 Strong or global supervenience assures that the evaluative area does not change, by fixing the same valence over all possible worlds.
.: So, weak supervenience does not really assure an adequate account of the dependency of evaluative upon the descriptive, whereas the strong supervenience does.
Ad Pr3: Notice that this premise really argues against particularism, and for generalism, by claiming that there cannot be any value change across all the possible worlds (if there is almost the same subvenient basis). But the argument presupposes that there is at least one change (in thick properties, or in physical properties), and this may lead to extensive valence changes in the area of evaluative. And this is just what the holistic particularism presupposes: That there may be an extensive change and also complete valence change of the area of evaluative based on slight changes in the valence base (the overriding).
Notice as well that Pr3 is utterly implausible. It is just not the case that evaluative valence properties do not change with small changes in the subvenient basis (may these be thick properties, or physical constituents of the subvenient basis). According to the particularism, this change is all pervading. And this particularist claim is well based on the experiences of how the valuation functions, whereas this is not the case with generalism.
What is the supervenience basis: Two candidates a. Thick properties (courageous, benevolent). b. Physical properties, constituents.
Presuppositions of generalism (i) atomism, (ii) tractability.
An argument for weak supervenience and against its negative treatment by generalism.
Pr1 Strong supervenience does not allow for any valence change in the evaluative properties, and therefore it is unrealistic.
Pr2 Weak supervenience allows for such change of valence involving evaluative properties.
.: So, weak supervenience assures an adequate account of dependency of the evaluative upon the descriptive, whereas strong supervenience does not.
Against presuppositions of generalism: (i) holism, (ii) nontractability.
Weak supervenience is unjustifiedly neglected. Especially, weak supervenience is appropriate for particularism, fitting to it. Why? Because weak supervenience allows for naturalism (no change in supervenient properties without the change in the subvenient basis); but weak supervenience also allows for weird behavior: just a small change in subvenient properties may have as the consequence complete reversal of supervenient (evaluative, moral) properties. This seems to be a big fault of the approach. But it is just a big fault from the point of view, from the perspective of generalism, which starts with the presupposition of tractability and of projectability.
Kim starts with the presupposition of projectability. This goes for him together with the presupposition of atomism, i.e. that there are atomistic elements or ingredients in the subvenient base, the elements (subvenient properties) that get tractably connected between them. This atomistic and tractable approach is directed against holism. But holism is characteristic for particularism. According to particularistic holism, there is no projectible pattern. And the holistic argument claims that what appear as slight continuous changes from the point of view of generalist projectible patterns are really bunch of singular cases. Each of these cases is intertwined with the rich holistic background of which it is a result. So what may appear as a general pattern linking several cases (that may be similar, yes) from the point of view of a generalist is just a wishful thinking unjustified projection. About projectability, which is grounded on similarity, one may look at Goodman’s thesis that the projectability is not justified: quus instead plus may arise (see also Kripke on Wittgenstein), and this depends on induction, where similarity just is not enough to secure the general pattern. This general pattern is more a wishful thinking along the generalist lines. Already induction is a problem for generalist, because it does not allow for general patterns to come through all the way. All these are generalist presuppositions. What needs to be done is to turn the perspective all around, and start with the presupoposition of rich holistic background supported structure for each particular occasion, with no general pattern extending over similar structures.
The claim is that weak supervenience is compatible with particularism. But as particularism is giving the right account of (moral) deliberation, so weak supervenience is compatible with the naturalistic account of (moral) deliberation.
Kim’s argument against weak supervenience is actually an argument for generalism and against particularism, in the sense that it defends general patterns and projectibility as precondition of supervenience. But it is far from clear that generalities are needed to assure the supervenience, or to assure naturalism, as for that matter. Supervenience and naturalism are well possible without generalities. Or, the issue of generalities is at least orthogonal to the issue of supervenience and naturalism.
The people before Kim, i.e. people like Moore and Hare were right: weak supervenience is all that is needed and it is also sufficient for what is needed. Strong supervenience would put too harsh requirement for materialism, especially if materialism would like to stay nonreductionist. (See Horgan’s position on this.)
The weak supervenience is not weird. It is just the right naturalist account of supervenient properties. There does need not be any projectability involved into the project of specifying supervenience. People did not wish to upgrade the project with any projectability. They did not wish to upgrade it to the strong supervenience. Weak supervenience sufficed for the task: specifying the naturalistic basis of supervenient properties.
Goodman, Nelson. (1973). Fact, Fiction and Forecast. Indianapolis: Bobbs-Merrill.
Horgan, Terence. (1993). “From Supervenience to Superdupervenience: Meeting the Demands of the Material World”. Mind 408: 555-586.
Horgan, Terence and Timmons, Mark. (1992). “Troubles on Moral Twin Earth: Moral Queerness Revived.” Synthese 92: 221-260.
Kim, Jaegwon (1993). “Concepts of Supervenience.” In Supervenience and Mind. Cambridge: Cambridge University Press.
Kripke, Saul A. (1982). Wittgenstein on Rules and Private Language. Oxford: Basil Blackwell.
Little, Margaret (2000). “Moral Generalities Revisited”. In Hooker B.W. and Little M. Moral Particularism. Oxford: Oxford University Press
Potrč, Matjaž (Unpublished). »Particularism and Productivity Argument.«
 This paper originated from a discussion with Noa Latham who pushed for the specification of supervenience relation in the transition from the descriptive to the evaluative properties, arguing for the plausibility of strong supervenience. Discussion with Vojko Strahovnik (Marijo Bilus joined in) brought me to the brink of formulating the defense of weak supervenience.