In The Mind and its Place in Nature, C.D. Broad tries to show, as he says (p. 59), that "there is no doubt" that the Theory of Emergence is a logically possible view with a good deal in its favor. And in his history of British Emergentism, McLaughlin states that emergentism is perfectly internally coherent, although he doesn't think it has any empirical evidence in its favor at present. I am inclined to agree with the assessment that emergentism is a coherent theory, but I can't see that Broad or McLaughlin has shown this, at least if this is taken to mean that it represents a metaphysical possibility (and not merely is not self-contradictory). Moreover, I suspect many people find the concept of emergent properties somehow unscientific, mysterious, and possibly incoherent; and I doubt that Broad has adequately addressed this intuition. He has indeed admirably explicated the meaning of the theory, so that now we can all understand it; but just as even to explain the idea, for example, that there are alterations without causes in such a way that it becomes absolutely, crystal clear what one means by "an uncaused change" would not be to demonstrate per se the metaphysical possibility of such things, Broad has not shown that there could be emergent properties. Because of this, what I would like to do in this paper is to try to articulate the implicit philosophical complaint that I think most people who find the idea unscientific have against the theory of emergence. I do not say that this complaint is valid; I merely want to identify it.
Now the rough idea behind emergentism seems to be the denial of the general principle that the whole can be fully explained in terms of the parts, or, to put the matter in more metaphysical terms, the idea that the whole is greater than the sum of the parts. Broad's formulation of emergentism sounds to me like an interpretation of this idea:
On the first form of the theory the characteristic behavior of the whole could not, even in theory, be deduced from the most complete knowledge of the behaviour of its components, taken separately or in other combinations, and of their proportions and arrangements in this whole. This alternative, which I have roughly outlined and shall soon discuss in detail, is what I understand by the "Theory of Emergence". (p. 59; emphasis in original)
Two pages later he redescribes the theory in almost the same words but this time substituting "properties" for "behaviour." Then he goes on to amend the formulation by allowing compositional principles in the would-be deduction base: he says that an emergent property has to be impossible to deduce from properties and relations of parts together with any simple compositional principle that could be discovered by examining things other than the particular kind of complex in question.
I have reasons for changing Broad's definition. A definition of emergence should at least satisfy these criteria: (1) It should describe a wholly metaphysical and not an epistemic relationship; (2) it should fit in with what emergentists say about emergent phenomena; (3) it should not be motivated by any unsound arguments.
Broad's formulation talks about deducibility from a certain kind of knowledge. Deducibility is an epistemic notion, and even with the phrase "even in theory," Broad's definition of emergence makes reference to observers. It is reasonable to suppose that what Broad meant was that emergence is a certain metaphysical relationship between a whole and its parts such that it is owing to the existence of this relationship that it would not be possible to deduce the properties of the former from the properties and relations of the latter. Non-deducibility is a symptom of emergence. Now what in the world kind of relationship could this be? the anti-emergentist wonders. Wouldn't it have to be, naturally, that in some metaphysical sense the former are not sufficient for the latter, that the facts about the whole are not contained in the facts about the parts but are something over and above them? This is the suggestion I shall adopt.
Broad also makes reference to observers when he characterizes the kind of compositional principle that is allowed in the deduction base for a property that isn't emergent - he characterizes the compositional principles as being discoverable by a certain means. I shall come back to this issue shortly.
Next, Broad doesn't say that when emergence occurs the properties that belong to a whole at a given time cannot be deduced from the properties and arrangement that the parts have at that time. He says that the properties of the whole can't be deduced from knowledge of the properties that the parts have when they aren't composing the whole and their arrangement in the whole. However, I think the emergentists really hold the former doctrine. In "Mind - Dust or Magic?", Van Cleve defends emergentism on the grounds that no properties of elementary particles could entail mental properties in people. He attacks Thomas Nagel's panpsychism on the ground that intelligence in a complex system not only won't follow from any physical properties of the parts but also won't follow from mental properties or proto-mental properties of the parts. He says that therefore even if we postulate mental or proto-mental properties in elementary particles we would still have to admit a kind of emergence. If this is an argument for emergence, then it certainly sounds as if emergence is the phenomenon in virtue of which properties of a whole at a given time do not logically follow from properties of the parts at that time. Van Cleve does not believe that when they get combined into a human being, elementary particles take on some new properties such that the existence of consciousness follows from the presence of these properties; rather, he is arguing that the properties of the elementary particles, whatever they might be at any time, cannot logically imply the existence of consciousness in the whole system. Most likely, he holds this on Cartesian grounds - i.e., you can always imagine a possible world in which any given physical phenomenon occurs without any mental phenomenon, or in which our particles are conscious and we're not, etc. This general consideration justifies the view that the properties of a human who is conscious at some time are not entailed by the properties of his parts at that time.
Third, Broad introduces the amendment about compositional principles following and because of his discussion of the fact that
[I]n no case could the behaviour of a whole composed of certain constituents be predicted merely from a knowledge of the properties of these constituents, taken separately, and of their proportions and arrangements in the particular complex under consideration. (p. 63; emphasis in original)
He wants to avoid the trivialization of the theory of emergence that would make even, e.g., mass, turn out to be emergent since to infer the mass of a whole from the mass of its constituents requires a contingent principle of the additivity of mass. But if we adopt the suggestion of the previous paragraph, this problem does not arise: I think that the masses that any constituents have at any point in time does logically imply the mass that the whole will have at that time. It is not conceptually possible, for example, that some objects should have masses of 4, 3, and 7 units but that the whole consisting of nothing but these objects should at the same time have a mass different from 14 units, although it may be possible that when some objects combine into a whole one or more of them change their masses, so that the mass of the whole would not be the sum of the masses the parts had before the combination. Broad's illustration is of a pair of taps that pour water into a tank. He says that the amount of water that both put into the tank per unit of time when simultaneously running is not necessarily the sum of the amounts each would put into the tank separately, because, for example, if the two taps came from the same pipe, less would flow from each when both were turned on together than when each was turned on separately. This, of course, is not, nor is it intended as, a counter-example to the principle that the total rate of flow of the two pipes at any time is the sum of their individual flow rates at that time.
Similarly, it seems necessary to me that if, say, two forces simultaneously act on a body without any other forces being involved, then the total force on the body is the vector sum of the individual forces at that time.
If this is correct, we can dispense with the amendment about compositional principles, and therewith also the worry about the epistemic characterization of them Broad gives. We now have that a property of a whole is emergent if the fact that the whole has that property at some time is something more than or 'over and above' the parts having the properties and arrangement they have at that time.
My sense is that the philosophical resistance to this theory stems ultimately from an (again, implicit) theory that is widely held today about, roughly speaking, the composition of the world. I like to think of it as the Atomistic-Subjectivist Theory of Composition. According to this theory, there are some basic constituents of the world, quarks, leptons, bosons, or whatever the physicists are going to tell us are the elementary particles, and these particles are something similar to natural kinds. Dividing the world up conceptually into elementary particles is the correct way of dividing it up because the world is in itself, objectively divided into elementary particle tokens. Each of these particles is a simple substance in the philosopher's sense: its existence is independent without qualification - that is, the fact of its existence does not depend on the existence of anything else, either in the sense in which the existence of a composite derives from the existence of its parts, or in the sense in which the existence of some quality depends on the existence of a subject to which the quality appertains, or in any other sense that may be imagined. All this may not be entirely clear, but this is the atomistic part of the theory.
The subjectivist part states that whether some collection of particles constitute a whole is purely subjective; it is a matter merely of convention or of our way of perceiving or thinking about the world. The whole-part relationship is metaphysically empty in a sense. For example, let me define "flurg" to denote my left eyeball plus the dark side of any planet in the solar system. The fact that we don't countenance flurgs as part of the furniture of the world but we do countenance cats does not reflect any fact about flurgs and cats that the latter exist and the former don't. Any collection of things can be an object. If there are n elementary particles in the universe, then there are exactly 2n - 1 possible objects (this being the number of distinct sets of elementary particles) that we could define, and each of these would be on all fours in terms of legitimate object-hood with cats. I don't think it matters much whether one expresses the idea by saying that there aren't really any composite wholes (Nihilism), by saying that any collection of particles is a whole (Universalism), or by saying that whether some collection of particles constitute a whole is subjective, that is, determined by facts about observers, external to the system of particles. All of these formulations stem from the same idea, I think, namely, that of the metaphysical emptiness of the concept of wholeness. (If you think that a concept is empty, you might conclude that therefore it can legitimately apply to anything, or that it really applies to nothing, or that it ipso facto applies to the things we apply it to.)
As I say, I think approximately this picture is present in the minds of many educated and at least mildly philosophical people today. Furthermore, omitting specification of just what the elementary particles are, I hypothesize that most of these people just can't see how the theory could fail to be true, and especially, they can't see how the rest of the theory could fail to be true if it is granted that the world is composed of elementary particles. And I think it is owing to that that they cannot see how there could possibly be emergent properties. I shall try to explain why presently.
Our atomist-subjectivist theory suggests that all facts should be constituted by, that is, should be necessarily implied by, the sum total of facts about elementary particles. Why? Because that's what there is, and that's all there is. And the facts about elementary particles consist entirely in the properties they have, their arrangement, and the relations they stand in to each other. So there shouldn't be any facts over and above these, as emergentism implies.
To put it another way, emergentism claims that 'the whole is greater than the sum of the parts,' but this is impossible because 'the whole' only exists in the eye of the beholder. What there really is are the elementary particles, and our taking some stance towards a certain set of them cannot alter their nature.
In reply to this objection, an emergentist might, while conceding that wholeness isn't an objective property, protest that his theory doesn't imply that wholeness per se makes a difference to the behavior or properties of any system of particles but that some other, objective relation in which these particles stand to each other makes a difference to their behavior. And that it makes the difference it does is just a brute nomological fact, so it doesn't have to be a logical consequence of the facts about the parts. (This last point is suggested by Van Cleve.)
This reply is probably inadequate, for the anti-emergentist claims, as I think the Atomistic-Subjectivist Theory of Composition indeed implies, that any property a system has is a logical consequence of the properties and relations of the parts, and any behaviour of a system is a logical consequence of the behaviour of the parts, though he need not claim that there is anything whose behaviour is a logical consequence of its properties. The problem with the suggestion that it's just a contingent causal law that certain complex systems have certain global properties isn't that we're causal rationalists. It can very well be the case that particles getting into a certain configuration has contingent effects. But the problem is that in this particular case, there is nowhere for the effects to reside. There isn't anything that the emergent phenomena could consist in if they don't consist in facts about elementary particles, given that nothing but elementary particles exists.
To take a familiar example, what would it be like if society exhibited some behavior that was not a logical consequence of the behavior of its members? Suppose a society goes to war: doesn't this just follow from a large number of its members going off to fight under direction of a certain other group of people in authority, etc.? If someone said, no, the war is something over and above the all of the individual fighting, what could he possibly have in mind - what else could be involved? Is there some possible world in which every individual behaves exactly the same way but the society behaves differently? I can't see what such a world would be like.
Broad discusses the vague sense that non-mechanistic theories are radically unscientific (pp. 73-81), but all he says in emergentism's defense is that chemistry and physiology are able to proceed perfectly well without assuming that mechanism is true. This isn't very satisfactory since it doesn't address the question whether, despite this, mechanism is supported by some known scientific facts, nor whether emergentism is coherent. There may be plenty of mysterious, unscientific, and incoherent ideas that scientists don't have to base their work on the falsity of.
I think these considerations are simple and intuitive. I'm not sure exactly how an emergentist ought to reply, but I suspect his best bet would be to give an alternative to the atomist-subjectivist theory. He should say either that it isn't an objective fact that the world is composed of elementary particles or that it is an objective fact that it also contains cats. I can't go into the details of either of these views here, however, both because I don't know them and because it is beyond the scope of this paper. I think in order to show that emergentism is coherent one needs to show that some such view is possible.