Emergence and Downward Causation

ion inherent in each single individual. In its reality it is the ensemble of social relations’ [Marx 1972: 109]). If one were to represent these two views of emergence as ideal types, one could say the former, complex systems approach stresses the ordinary workings of nature as resulting in surprising behaviour of complex systems; the distinction between levels is characterized as a distinction between larger and smaller, or between mereologically composed wholes and their parts, with an emphasis on mundane processes, complicated only because they contain elements of feedback, producing ‘resultant’ emergence. This contains both a diachronic and synchronic aspect: the later, evolved system contains more than its simpler predecessors, and the behaviour of the later system is more complex with respect to its simpler parts and cannot be reduced to the behaviour of those parts. The second, holistic and relational view is less concerned with what may be called ‘unadorned complexity’. In the past, adherents of holism were impressed with the difference that life makes, and so were apt to stress the intricate interconnectedness and interdependence of what may be called ‘vital’ properties. On this view the difference of levels is not only one of larger complexes having properties not reducible to the properties of their simpler parts. Although such irreducibility is endorsed, there is an attempt to make it more substantial, or more Donald chap10.tex V1 November 20, 2009 1:45pm Page 142 142 Cynthia Macdonald and Graham Macdonald principled. In the nineteenth century this produced the doctrine of ‘vitalism’, a view insisting on the uniqueness of the properties that constituted the essence of living forms. The difference of levels was seen as a difference induced not just by the emergence of new properties, but of a new type of property. As a consequence some adherents of this view were less inclined to explain this ‘emergence’ as being the result of simple processes working in unforeseen ways; such an explanation would have been seen as unduly materialist, relying on mechanistic causality which, it was claimed, could never produce the inbuilt telos of vital properties and processes. In general this approach emphasized synchronic emergence, the temporal origin of uniqueness being somewhat difficult to explain without invoking a deus ex machina. Perhaps for this reason the holistic view of emergence is not particularly popular today. Even before Darwin and Wallace naturalized biological complexity, and teleology, via natural selection there was a group of scientists in Germany, known as the Munich materialists, who did their best to discredit what they took to be an anti-scientific lobby constituted by Driesch and other vitalists, and did so by insisting that one could create life in a test tube by mechanical means. Combined with the growing influence of Darwin, the more extreme versions of biological holism gradually withered and died. Nevertheless, we want to keep this view on the table, since it contains, we think, elements of our topic that are crucial to the conception of emergence we want, and think appropriate here, to pursue. In what follows, we won’t have much to say about what we call the ‘complexity thesis’, mainly because many of the issues here seem to us to be empirical. Our suspicion is that many of the enthusiasts for emergence conflate surprise, unpredictability, and (present) inexplicability with ‘in principle’ inexplicability. It also seems, to our sceptical eyes, that not even present inexplicability is deemed necessary for this type of emergence to exist. The behaviour of many complex systems is not inexplicable, though it can be surprising. The discovery that the cells composing multicellular slime molds do not have a ‘leader’, a cell assuming the organizing role, was surprising, given that the behaviour of the multitude of cells seemed to call for such an organizer. But we do now have explanations of how the cells composing slime molds behave the way they do, and why the price of coffee in Starbucks is what it is, and our view is that this type of ‘emergent’ behaviour does not present any serious concern for the philosopher. The difference between complex systems emergence and holistic emergence has been characterized by David Chalmers (2006) as a distinction between weak and strong emergence. This distinction has been remarked upon by a number of philosophers and scientists; Herbert Simon, for example, claims that weak emergence is the view that the parts of a complex system have relations that do not exist for the parts in isolation. For example, the template of an enzyme has no function until it is placed in an environment of other molecules of a certain kind. Even though the template’s function is ‘emergent’, having no meaning for the isolated enzyme molecule, Donald chap10.tex V1 November 20, 2009 1:45pm Page 143 Emergence and Downward Causation 143 the binding process, and the forces employed in it, can be given a wholly reductionist explanation in terms of the known physico-chemical properties of the molecules that participate in it. (Simon 1996: 170–1) Like us, Chalmers thinks that weak emergence contains little of philosophical interest: If one is given only the basic rules governing a cellular automaton, then the formation of complex high-level patterns . . . may well be unexpected, so these patterns are weakly emergent. But the formation of these patterns is straightforwardly deducible from the rules (and initial conditions), so these patterns are not strongly emergent. . . . The existence of unexpected phenomena in complex biological systems, for example, does not on its own threaten the completeness of the catalogue of fundamental laws found in physics. (Chalmers 2006: 245) This leaves us with strong emergence, of which there have again been many characterizations. Chalmers has this to say about it: We can say that a high-level phenomenon is strongly emergent with respect to a low-level domain . . . when truths concerning that [high-level] phenomenon are not deducible even in principle from truths in the low-level domain. (Chalmers 2006: 244) This is clearly a characterization of a synchronic relation; it is not part of this view that if one cannot deduce truths about the later stages of a process from truths about the earlier stages, then the later stage is strongly emergent. Any output of a non-deterministic process would qualify as strongly emergent if we were to use the above criterion diachronically, and that is clearly not the intended interpretation of those wishing to espouse strong emergence. Now, we think that this is a very strong version of strong emergence indeed: pack enough into descriptions of the lower domain, including its history, and, arguably, truths about biofunctional properties will be deducible, and so not emergent. It is unsurprising, then, that the only candidate for strong emergence countenanced by Chalmers is consciousness, the notorious explanatory gap ensuring non-deducibility. The rest of the psychological and social world, it is claimed, will be emergent with respect to the physical, but that emergence will be dependent on the physical facts plus consciousness, and so is not intrinsically emergent. Given the truths about the physical and the truths about consciousness, the truths of the social, say, will be deducible. The plausibility of this extreme view will depend to some extent on how it is fleshed out; we need more detail about which truths are included as truths about consciousness in order to assess the claim that given all those truths plus the physical truths, all the rest of the truths about the world are deducible, including truths about the price of Starbucks coffee. We will have no chance of deducing truths about the latter type of fact unless truths about representational facts, about propositional attitudes, are included as truths about consciousness—or so we believe. Donald chap10.tex V1 November 20, 2009 1:45pm Page 144 144 Cynthia Macdonald and Graham Macdonald Further, there is another famous explanatory gap that is not considered by Chalmers in this context, and that is one constituted by the is–ought gap, or, more generally, the supposed fact–value dichotomy. If there is the notorious Humean gap here, it would appear that there are two options available with respect to moral truths: deny there are any, or admit another strongly emergent domain. One reason for denying there are any is important to our present topic, though it has not been discussed directly in relation to it. This concerns claims about the causal inertness of moral properties (Harman 1977, 1985, 1986); that it is not instances of moral properties, but rather, of our beliefs about moral truths, that cause us to behave the way we do, the moral property being at best epiphenomenal. The lack of any causal power is thought to be a sure sign that there is no real moral property doing any causal, and so any explanatory, work. If this is right, then non-deducibility will need to be supplemented by a more substantial test for emergence, given that emergent properties are claimed to have genuine causal powers. The formulation of strong emergence in terms of the relation between the truths about different domains is what we would call a ‘formal’ notion of emergence. While the non-deducibility of the higher-level truths is clearly of interest, one would like to know more about why the nondeducibility holds. What is it about the nature of phenomena and/or properties in the relevant domains that makes for the in principle non-deducibility of the truths of one domain from the truths of the other? Leaving it as a brute fact is unsatisfactory; it invites mystery where there should be none. Our own preference is for a more straightforwardly metaphysical interpretation, one which talks of emergent properties, rather than non-deducible truths, but of course one then needs to say what it is about the properties that makes them emergent. And, again, the problem is that simply asserting their irreducibility to physical properties is unsatisfactory; one needs an argument as to why irreducibility holds. We think that the argument will be different for different cases (Macdonald 1992; Macdonald and Macdonald 1995). If one wants to defend the irreducibility of biofunctional properties, then one will need to pay attention to what kind of properties these are, note that they have a historical dimension, and that two instances of the same physical-chemical property may be different with respect to whether they are also instances of a specific biofunctional property. If they are thus different, then that physical-chemical property cannot be identified with the biofunctional property with which it sometimes shares an instance, it being a condition on such an identity of properties that they necessarily share all their instances. If one wants to defend the irreducibility of mental properties, then one will need to mount a different argument, one specific to the nature of the mental; and it may need to be different with respect to different types of mental property as well. An argument to the effect that intentional properties are irreducible will be likely to take a different form from an argument to the effect Donald chap10.tex V1 November 20, 2009 1:45pm Page 145 Emergence and Downward Causation 145 that experiential properties are irreducible. And so on, for other cases, say, that of moral properties. We do not propose to mount any of these arguments here,2 since we think that FN:2 prior work needs to be done: there are powerful arguments purporting to show that irreducibility has principled problems, and this is what the present discussion is aimed at tackling. These arguments hark back to our brief discussion of moral properties, and to the thought that, when faced with the non-deducibility of truths about a domain from the truths about a lower-level domain, when faced with an explanatory gap, one has a choice. To put it succinctly, one can be emergentist or eliminativist, be realist or irrealist with respect to the relevant property-type. Or one can deny irreducibility and non-deducibility. Which way one goes will be case-dependent, but as we have indicated and will shortly discuss in more detail, considerations about causality will be highly relevant. Given that Chalmers’ version of strong emergence is too strong for our taste, and given our preference for a more robustly metaphysical version of the doctrine that focuses on properties rather than on truths, here is our favoured version of strong emergence suitable for the likes of non-reductive monism (i.e., for a physicalist position on the nature of mind): A property, M , is an emergent property of a (mereologically complex) entity/event, e, if