Search Results

This chapter explores the relationship between reductionism and the mechanisms approach to extrapolation. It maintains that the mechanisms approach is committed to reductionism only insofar as it requires that micro-mechanisms are typically correctively asymmetric with regard to macro-level generalizations, and that corrective asymmetry in this sense is consistent with pluralism.

This chapter provides an account of realization within a mechanistic framework and introduces the notions of variable realizability, multiple realizability, and medium independence. Realization is the relation between a higher-level property and the lower-level properties of which it is an aspect. Variable realizability occurs when the same higher-level property can be realized by different lower-level properties—different lower-level properties share the same aspect. Variable realizability is ubiquitous yet insufficient for multiple realizability proper. Multiple realizability proper occurs when the same higher-level property can be realized by different lower-level properties that constitute different mechanisms for that property at the immediately lower mechanistic level. Medium independence is an even stronger condition than multiple realizability: it occurs when not only is a higher-level property multiply realizable; in addition, the inputs and outputs that define the higher-level property are also multiply realizable. Thus, all that matters to defining a medium-independent higher-level property is the manipulation of certain degrees of freedom. Medium independence entails multiple realizability, which in turn entails variable realizability, but variable realizability does not entail multiple realizability, which in turn does not entail medium independence.

This chapter argues that the skeptical argument impugning psychophysical identities is significant if convincing and that the significant impact of the argument comes at little cost, in terms of controversial philosophical resources. The chapter begins by clarifying the conclusion of the skeptical argument: namely, that we cannot be warranted in accepting any specific psychophysical identity statement. The chapter then argues that the skeptical argument is significant for functionalists and multiple-realizability theorists, who often appeal to supervenience or token identity, even though the skeptical argument concerns type identity; and that anyway the argument can be redirected to address supervenience claims and assertions of token identity. The chapter accepts a cost that eliminative materialists are unwilling to pay and argues that in practice, analytic functionalists tend to embrace a form of eliminativism. The chapter does not commit to some claims associated with Chalmers, including two-dimensionalism, modal rationalism, and apriority.

Block and Fodor classically offered three lines of indirect evidence for multiple realization: The Lashleyan doctrine of equipotentiality, evolutionary convergence on psychological traits, and the conceptual possibility of artificial intelligence. However, evidence of the first sort falls short of demonstrating that different kinds of brains can produce the same psychological functions by instantiating relevantly different neural properties. Evidence involving evolutionary convergence runs the risk of confusing convergence in behavior with convergence in psychological processing; and even if this difficulty is avoided, the presence of evolutionary constraints suggests that convergent psychological traits may well have convergent realizers as well. Finally, the conceptual possibility of artificial intelligence begs the question in favor of multiple realization. Given that no such intelligences exist today, no conclusions about the nature of their hypothetical realizers follow.

This chapter takes Chapter 5's causal exclusion argument to pose a dilemma about the reality of multiple realisable properties: either they are reducible to first-order physical properties or they are not associated with distinctive causal powers. It detects two strands in Chapter 5's challenge. The Incoherence Challenge is that it is incoherent to hold that one and the same set of objects or events is anomic at one level of description, but nomic at a different level of description. The Conventionality Challenge is that nomicity should depend on objective similarity, and not merely on how things are described. It is argued that both challenges can be met, so that we can find a third way between the horns of Chapter 5's dilemma, and vindicate multiple realisability.

Many arguments for multiple realization begin with the claim that psychological states are computational states, and conclude the abstractness of computational states assures their multiple realizability. However, in many cases where a cognitive scientist offers a computational description of a cognitive capacity, on offer is merely a computational gloss that carries with it no commitment to the existence of an internal psychological organization that actually performs computations. Similarly, that the abstractness of computational descriptions entails the multiple realizability of psychological states runs into difficulty. Abstractness is a property of descriptions, not of the processes described. Hence, the abstract character of computational descriptions does little to support the claim that the processes they describe might be multiply realizable. Furthermore, one may insist that computational descriptions of psychological processes succeed because the processes themselves are computational, but this response neglects decades of research that challenges the computational nature of psychological processing.

Bradley and James’s disagreement as to how the explanation of attention should proceed can be traced to a disagreement about the metaphysical category in which attention belongs. This chapter gives an account of the distinction between their two rival views of attention’s metaphysical category. James assigns attention to the ‘process-first’ category. Bradley assigns it to the ‘adverbial’ category. The distinction is explained by deriving it from a more fundamental distinction between those taxonomies that classify events on the basis of properties had by the events themselves and those taxonomies that classify events on the basis of properties of the objects that participate in those events. This account enables us to see that Bradley’s claim was distinct from (and explanatorily more consequential than) more recent claims about multiple realizability, natural kinds, and levels of explanation. The chapter concludes by relating this distinction to different claims about attention’s supervenience base.

There are reasons to look at two existing kinds of theories of consciousness, including the need to elicit criteria by which to judge any theory of consciousness. One kind is dualism, still persisting, and the other is functionalism, dominant in the sciences of mind and consciousness. They are treated differently, wrongly. Dualisms are taken to be to the effect that consciousness is non-physical, taken so persuasively if without the benefit of an understanding of physicality. Abstract functionalism is to the effect, unnoticed, that being conscious, whatever its relation to brains or computers, is unphysical. This is the upshot of several prized propositions, one about multiple realizability of consciousness in physical bases. Abstract functionalism is therefore as vulnerable to certain objections as the dualisms. Does much of science in fact tacitly contemplate physical functionalism, which is true to its name? A pity if so, since physical functionalism as naturally understood is self-contradictory.

This chapter presents and defends a positive position on multiple realizability. It is quite liberal in that any multiplicity of realizations counts as multiple realizability. This is palatable once we accept that multiple realizability claims, like reduction itself, are always relative to a level of abstraction. The theoretical options that arise whenever there is apparent realization-level heterogeneity are explored. But the account is also quite strict in that it requires exact sameness at the level of the super-determinate, realized kind. The account is applied to particular examples—such as memory and computer science—to illustrate the differences between multiple determination and multiple realization. The various senses of autonomy, as discussed in Chapter 4, are then derived from this account. Finally, the nuanced connections among singular realization, singular realizability, and identity are explored.

The idea of multiple realization begins with a simple thought: There exists sameness of kind at one level of description, and differences in kind at a lower level of description. Two organisms, for instance, might be described as sharing a perceptual state, while differing in the physical states that realize this state. When investigating the possibility of multiply realized psychological states, one cannot confuse psychological similarity with behavioral similarity, as one might when investigating whether a capacity like face recognition might be multiply realized. Two kinds of organism might be able to recognize faces, but only one (or neither!) might do so through psychological means. The shared behavior is thus not evidence of a multiply realized psychological process. Finally, modal intuitions regarding the possibility of multiple realization do not come for free. They should be grounded in what our best sciences tell us about the prospects for actual multiple realization.

In his book, Truth as One and Many, Michael Lynch presents a theory of truth with two main components. The first is a functionalist account of truth: a proposition is true just when it has some property that plays a certain characteristic role, the truth-role. The second is the thesis that truth is multiply realizable: truth can be realized by different properties for propositions belonging to different domains of discourse. This chapter looks at the structure of Lynch’s functionalist account of truth while keeping a close eye on his multiple-realizability thesis. It is not a straightforward matter to understand how the two components of his theory fit together. The main aim of the chapter is to provide constructive criticism and clarification. It scrutinizes relevant aspects of Lynch’s view in considerable detail, hoping that this will tell us something interesting about functionalism concerning truth and maybe even about functionalism in general.

This chapter offers criticisms of many views of multiple realizability that do not appropriately distinguish it from cases of multiple determination. The basic idea of multiple realization is sameness of realized kind through differences in realizing kinds. So, the distinction between specification differences and differences in level of abstraction (e.g. determination dimensions) is central once again. Necessary and sufficient conditions for both core and total realization are defended. Multiple realizability is then connected to three different senses of reduction and autonomy: ontological, explanatory, and methodological. This view of multiple realizability is contrasted with rival accounts—by Stephen Yablo, Sidney Shoemaker, Lawrence Shapiro, and Jaegwon Kim, among others—and argued to be superior.

Proponents of the idea that reality is hierarchical appeal to the ‘multiple realizability’ of higher‐level properties. The ontology of multiple realizability is fuzzy, however, and in any case, we can accommodate putative examples of multiple realizability without positing higher‐level properties. Predicates taken to name such properties are better understood as being satisfied by diverse but similar properties. Ontological reduction does not imply analytical or explanatory reduction.

This chapter relates the philosophical concept of multiple realizability to the physics concept of universality. It discusses and responds to Elliott Sober’s defense of reductionism in the face of multiple realizability. Further, it introduces an important explanatory question (labelled AUT). This asks how systems that are heterogeneous at some micro-scale can exhibit the same pattern of behavior at the macro-scale. It is shown that reductionists do not have the resources to provide a successful answer. Two, related, answers are proposed. One involving Renormalization Group arguments, the other invoking the theory of homogenization.

The systems studied in the special sciences are often said to be causally autonomous, in the sense that their higher‐level properties have causal powers that are independent of those of their more basic physical properties. This chapter aims to clarify what is implied by the doctrine of the causal autonomy of special‐science properties and to defend the doctrine using an interventionist theory of causation. In terms of this theory, it shows that a special‐science property can make a difference to some effect while the physical property that realizes it does not. Moreover, the theory permits identification of necessary and sufficient conditions for the causal autonomy of a higher‐level property, and to show that these are satisfied when causal claims about higher‐level properties have a special feature we call realization‐insensitivity. This feature consists in the fact that the relevant claims are true regardless of the way the higher‐level properties they describe are physically realized. The findings here are consistent with those of other philosophers, for example Alan Garfinkel, who have noted the realization‐insensitivity of higher‐level causal relations as a distinctive feature of the special sciences and have suggested that this feature ensures their independence from lower‐level causal relations.

There is an ongoing philosophical and scientific debate concerning the nature of computational explanation in the neurosciences. Recently, some have cited modeling work involving so-called canonical neural computations—standard computational modules that apply the same fundamental operations across multiple brain areas—as evidence that computational neuroscientists sometimes employ a distinctive explanatory scheme from that of mechanistic explanation. Because these neural computations can rely on diverse circuits and mechanisms, modeling the underlying mechanisms is supposed to be of limited explanatory value. I argue that these conclusions about computational explanations in neuroscience are mistaken, and rest upon a number of confusions about the proper scope of mechanistic explanation and the relevance of multiple realizability considerations. Once these confusions are resolved, the mechanistic character of computational explanations can once again be appreciated.

This chapter first discusses a tension in Fodor's treatment of special sciences. It argues that, on Fodor's account, special science counterfactuals are necessitated by fundamental physical laws and facts. So, if there are metaphysically independent special science laws then they can only overdetermine counterfactuals, which is puzzling. Next the lawfulness of special science regularities is considered. After examining Boltzmann's reconciliation of laws of thermodynamics with fundamental dynamical laws, the chapter proposes that the lawfulness of such regularities is grounded in the dynamical laws plus a probabilistic constraint on the initial conditions of the universe. It shows that physics misses no nomological/explanatory structure that the special sciences supply. This reductionist account denies the existence of metaphysically independent special science laws, but it does not entail that special science properties are identical to properties of fundamental physics and it allows for the multiple realisability of special science laws.