Philosophy of Mind

PHIL 4300/5300

Fall 2002

Bob Hanna

Handout #5

Supervenience and Emergence

(A) Supervenience

The basic idea behind the notion of supervenience is that it captures a modal dependency relation between types of properties that is somewhat weaker than identity, hence consistent with the denial of identity between properties of the relevant types--hence consistent with "property dualism" of some sort. So we can separate properties into two distinct classes: the lower-level or more basic properties, and the higher-level or less basic properties. Call the lower-level properties "A-properties" and the higher-level properties "B-properties." Then B-properties supervene on A-properties if and only if

(1) necessarily anything that has some property G among the B-properties also has some property F among the A-properties (or alternatively: no two things can share all their A-properties in common unless they also share all their B-properties in common) (or again alternatively: no two things can differ in any of their B-properties without also having a corresponding difference among their A-properties), and (2) necessarily anything’s having F is sufficient for its also having G.

This two-part supervenience relation is also what Jaegwon Kim aptly calls "strong supervenience," because we can define a weaker relation, "weak supervenience," by asserting feature (1) without feature (2). According to weak supervenience, there is no B-property difference without an A-property difference. The difference between weak and strong supervenience is that strong supervenience implies the existential dependence of B-properties on A-properties, whereas weak supervenience does not. So the relation of weak supervenience is consistent with the existence of possible worlds in which the A-properties exist but the B-properties do not.

Now back to strong supervenience, or just plain old "supervenience" for short. If we assume that the A-properties are physical properties and that the B-properties are mental properties, then this yields a materialist or physicalist supervenience. In this context, feature (1) of supervenience is known as the "necessary covariation" of the mental with the physical, and feature (2) is known as the "upwards dependence" of the mental on the physical. It is sometimes held that a materialist supervenience must add the proviso that feature (1) and (2) are further constrained by upwards nomological entailments running between the A-properties and the B-properties. When this extra constraint is added, materialist supervenience is then called "superdupervenience," because it captures the idea that the lower-level properties necessarily determine the higher-level properties in a wholly lawlike fashion.

In any case, it should be noted that the asymmetric modal dependencies involved in materialist identity, supervenience, and superdupervenience alike imply a strict non-reciprocity between the mental and the physical, since what is being asserted is not merely the necessary involvement of the mental with the physical but also the one-way or exclusively upwards dependence of the mental on the physical.

Now finally we can define logical supervenience. The notion of logical supervenience means that the two occurrences of ‘necessarily’ in the formulation of strong supervenience are to be read as ‘logically (i.e., analytically, weakly metaphysically) necessarily’, as opposed, e.g., to ‘non-logically (i.e., non-analytically, synthetically, strongly metaphysically) necessarily’ or ‘physically (i.e., nomologically, naturally) necessary’, which are more restricted modalities. As David Chalmers has pointed out in chapter 2 of The Conscious Mind, the importance of logical supervenience is its equivalence with the notion of reductive explanation. If B-properties logically supervene on A-properties, then B-properties follow analytically from A-properties and thereby exhaustively explain those properties, because a suitably well-informed and rational thinker could, merely from her knowledge of the A-properties together with her knowledge of any nomological entailments that hold between the A-properties and the B-properties, infer the B-properties. Thus another way of describing the logical supervenience of B-properties on A-properties is to say that the B-properties are "nothing but" or "nothing over and above" the A-properties.

(B) Emergence

The root or core notion of emergence is fairly simple: some wholes or systems have "global" properties that are neither reducible to (i.e., neither identical to nor logically supervenient on) the "local" properties of their proper parts nor deducible from the knowledge of those local properties. Global properties of this sort are emergent properties of those wholes or systems. Nevertheless and unfortunately for ease of philosophical comprehension, emergence has at least three different basic varieties.

According to the first notion (call it "emergence1"), emergence says that the global properties of some wholes or systems are neither reducible to the intrinsic properties of their proper parts nor deducible from the knowledge of those local properties. For instance, the global properties of the chemical compound hydrogen dioxide or H2O--i.e., the macroscopic properties of water--are neither reducible to the intrinsic properties of hydrogen (H) and oxygen (O) nor could they have been deduced from the knowledge of those local properties alone. This is in contrast to wholes or systems that are nothing but mere collections of their proper parts and thereby reducible to the intrinsic properties of their proper parts. Unordered sets or classes, for example, are uniquely determined by their membership. The significance of emergence1, then, is that it locates emergence in non-aggregative wholes.

Notably, emergence1 leaves open the possibility that the global properties of wholes or systems are logically supervenient on (although not identical to) the intrinsic and relational properties of their proper parts. So emergence1 is consistent with an explanatory and ontological reduction of global properties to local intrinsic and relational properties. This is mereological supervenience. For example, if scientific essentialism is correct, then the global "watery" properties of H2O are mereologically supervenient on the local intrinsic and relational properties of H and O.

But there is an important ambiguity in the notion of the relational properties of the proper parts of a whole. There is a trivial sense in which every emergent property of a whole is also a relational property of its proper parts. For the very idea of emergence is that an emergent property of a whole is a property that comes to be instantiated solely due to the holistic interrelation of its proper parts. So in that sense emergence vacuously entails mereological supervenience.

Nevertheless it is possible to separate the relational properties of the proper parts of a given whole W into two distinct classes:

(1) the relational properties of the proper parts of W in W itself (we will call these the "internal" relational properties), and (2) the relational properties of the proper parts of W in wholes other than W (we will call these the "external" relational properties).

For example, the internal relational properties of the proper parts of H2O (i.e., H and O) will obviously include its global watery properties. But the external relational properties of H and O, that is the relational properties of H and O in wholes other than H2O--e.g., in H2SO4 or sulphuric acid--will not include the global watery properties of H2O nor could the global watery properties of H2O have been deduced from knowledge of the external relational properties of H and O. In other words, the global watery properties of H2O are neither logically supervenient on the intrinsic and external relational properties of its proper parts nor are they deducible from the knowledge of those properties. So mereological supervenience fails when the relational properties of the proper parts of wholes are only external relational properties.

This leads to a second notion of emergence ("emergence2"). According to emergence2, the global properties of some wholes or systems are neither reducible to the intrinsic and external relational properties of their proper parts nor are they deducible from the knowledge of those properties. Chemical compounds, complex dynamic systems (e.g., the weather), and biological wholes (e.g., living organisms) all exemplify emergence2 as well as emergence1. This is contrast to various arithmetic wholes (e.g., the set of natural numbers) and mechanical dynamic wholes (e.g., coke machines or automobiles) that are mereologically supervenient on the intrinsic properties of their proper parts together with their external relational properties. Indeed, wholes or systems whose global properties are strictly Turing-computable functions, recursive functions, or more generally linear functions of their proper parts are all emergent1 but not emergent2. The significance of emergence2, then, is that it locates emergence in non-mechanistic wholes.

There is one further important aspect of emergence. Both emergence1 and emergence2 leave open the possibility that the global properties of wholes have no causally efficacious impact on the intrinsic causal powers of their proper parts. The causal impact of global properties of wholes upon the intrinsic causal powers of their proper parts is called "downwards causation" or "global-to-local causation." For example, the global or watery properties of H2O have no downwards or global-to-local causal impact upon the intrinsic causal powers of H and O in the compound H2O, since the causal powers of water just are the result of combining the intrinsic causal powers of H and O in the compound hydrogen dioxide. This also holds true of all aggregative and mechanistic wholes.

But matters appear to be different in complex dynamic systems generally and biological systems more specifically, especially those involving living sentient organisms. For instance, an animal’s global property of hunger seems to be able to mobilize or enslave its component limbs, organs, cells, and finally microphysical particles in such a way as to get them all over to where the food is. This leads to a third notion of emergence ("emergence3"). According to emergence3, (i) the global properties of some wholes or systems are neither reducible to the intrinsic and extrinsic relational properties of their proper parts nor deducible from the knowledge of those properties, and (ii) these global properties have an efficacious downwards or global-to-local impact on the intrinsic causal powers of their proper parts. Emergence3 is clearly the most robust form of emergence. So emergence3 is strong emergence. Correspondingly then, we can think of emergence1 as "weak emergence," and emergence2 as "moderate emergence." In any case the significance of emergence3 is that it locates emergence in self-organizing wholes.