Jaegwon Kim, Supervenience and Mind: Selected Philosophical Essays (Cambridge: Cambridge University Press), 1993.
If there were just one name associated with contemporary discussion of supervenience it would be Jaegwon Kim's, and it is gratifying to find so many of his papers on this and related topics united here essentially unchanged. Three of the 16 papers appeared in the mid 1970s, the rest just since 1982. They are divided into two Parts: one on metaphysical issues of events, causation and supervenience; the other on issues in the metaphysics of mind, especially mind-body supervenience and mental causation. Each Part ends with a valuable section of postscripts -- clarifications, extensions, second thoughts, suggestions for future research. The whole is prefaced by a précis of the papers and their inter-relations.
The essays in Part I "all have something to do with the relationships of determination and dependence ... holding for properties, events, and states of the world" (xi). They are motivated by a "realist stance on explanation," according to which the explanation say of the mental by the physical "must be grounded in some objective relation of dependence or determination holding for the explanans and the explanandum" (xii) Those in Part II mostly address "the problem of delineating the place of mind in a physical world" (xv). The problem is defined, and its possible solutions constrained, by two basic assumptions: physicalism, according to which "the world is fundamentally a physical world governed by physical law"; and mental realism, according to which mentality is a real feature and must therefore "have genuine causal powers ... to affect other events and processes of this world" (xv).
No brief review can do justice to this feast. Instead, in the spirit of loyal criticism, I'll concentrate on just the following. (For needed additional argument see Post (1995) [and (1999)], which contain further references.) Start with Kim's assumption that something is a real feature only if it has causal powers. In accord with this, he adopts the principle of the Causal Individuation of Kinds (CIK): objects and events fall under a real feature or kind or property just insofar as they have similar causal powers (326). CIK is a major premise in his argument that nonreductive physicalism entails a mysterious sort of downward causation "no underlying physical-biological properties can deliver" (350).
But nothing essential to nonreductive physicalism implies CIK, and in any case CIK is hit by counterexamples from biology. Consider two objects that fall under the biological kind "heart," indeed under the species-specific kind "human heart." Hearts have the proper function of pumping blood; to be a human heart is to be supposed to pump blood in humans. By Larry Wright's account of proper function, endorsed by Wesley Salmon and improved by Ruth Millikan, an object x has this function in virtue of a natural-selective history in which past hearts (or enough of them), by pumping blood, enabled the containing organism to reproduce at higher rates, this prior successful performance thereby causing in due course the production of x itself.
It follows that whether x belongs to the kind "human heart" is determined not by x's actual causal powers or mechanism, but by a natural-selective history in which x is a descendant. It follows further that a member of the kind can be so radically defective -- through trauma, disease, genetic defect -- as totally to lack the normal causal powers (those it is supposed to have). Thus two objects -- one radically defective, the other normal -- can fall under the kind yet have quite dissimilar causal powers, contrary to CIK. Indeed any proper-function kind -- hearts, eyes, brains, signal-sequences of proteins -- is not so individuated that its tokens fall under the kind just insofar as they have similar causal powers. This includes a number of mental features, like memory and belief, which are proper-function kinds. Related difficulties afflict Physical Realization (PR), according to which a system s can have a mental property M only if s enjoys a causal mechanism for the implementation of M in s (pp. 343-4).
Kim's whole approach is strongly individualist, not just CIK and PR. All a thing's properties are supposed to be determined by or supervenient on its own causal or other physical properties. This is explicit in Weak Supervenience (WS), which he deems "highly plausible to regard ... as minimally necessary for any claim of determination or dependency between sets of properties" (84). WS says that necessarily, for any x and any y, if x and y have the same physical properties P, then they have the same nonphysical properties N (pp. 58, 79). To meet various counterexamples, Kim lets the P-properties include relational P-properties. A thing's properties all supervene on, and are determined by, its own physical properties and relations.
But consider. Among the proper functions a device or behavior may have is that of being supposed to map onto some affair in the world. A certain honey-bee dance may be supposed to map onto nectar that lies in a specific direction-and-distance, and the bits of nectar and pollen that adhere to the dancing foragers have as one of their functions to tell other bees about the kind of nectar -- say peach-blossom. Thus an aspect of the dance complex is supposed to map onto the affair of the nectar's being peach-blossom. It does so map if and only if the nectar is peach-blossom.
Now assume one of the bees is ill with pesticide. As a result she visits peach blossoms west of the hive, returns with peach-blossom nectar adhering, but tokens a dance for a location to the east. Assume further that peach-blossom nectar happens to exist at this location to the east, so that the dance complex does map onto this nectar's being peach-blossom; but the location to the east is a mile downwind on the far side of a high hill where no bee has actually been, and indeed not only is there no physical relation between nectar and hive that relevantly affects the bees, there is none that does any work in determining that the complex maps onto this nectar's being peach-blossom. (The objection that there must be a physical relational property of the complex which does work in determining the matter is met by showing how the plausible candidate relational properties all fail to do so.)
In this kind of case, a nonphysical property (of mapping onto the nectar's being peach-blossom) is determined not by the complex's own physical properties and relations, but only by these together with the physical conditions that underlie the nectar's being peach-blossom. Two complexes can be the same as regards their physical properties and relations yet not the same as regards all their nonphysical properties, contrary to the individualist WS. The same is true of certain cases of true belief, which parallel the bee-dance case. Physicalists need therefore to replace any supervenience principles that entail WS with principles of supervenient determination according to which x's properties are determined by physical conditions, just not always by physical conditions that amount to physical properties or relations only of x. One such principle is the "global" supervenient determination Kim rejects -- wrongly, I've elsewhere argued, but that's another story. Meanwhile, it is one of the many merits of Kim's work that here as elsewhere it helps us to see so clearly what is at issue.
Post, J. F. (1995), "'Global' Supervenient Determination: Too Permissive?" in T. Savellos and Ü. Yalçin, (eds.), Essays on Supervenience. Cambridge: Cambridge University Press.