R. Keith Sawyer: Social Emergence. Societies as Complex Systems. Cambridge University Press. 2005. This is an ambitious book. Keith Sawyer attempts to show that his new emergence paradigm provides a means to resolve many of the central problems in the philosophy of the social sciences. He argues that various debates within the social sciences have lacked the conceptual resources to deal with emergence, and that these resources are now available in the new emergence paradigm. For example, conceptual problems with Vygotskian psychology (Chapter 7), Durkheim s social ontology (Chapter 6), Parsons social systems theory (Chapter 2), Giddens s theory of structuration (Chapter 7), and Bhaskar s critical realism (Chapter 5) will be dissolved with the help of ideas from complexity theory and the recent philosophy of mind. Even the perennial debates about methodological individualism will finally come to end in a manner that is fruitful from the point of view of empirical research. The new paradigm will also restore sociology to its rightful place as the foundational social science. What is this new emergence paradigm? According to Sawyer, it is a positivist, objectivist, scientific approach, and consequently it rejects subjectivism and interpretivism (231). It represents the third wave of social systems theory, adding the ideas of micro-level agent communication and social emergence to Parsons structural functionalism (the first wave) and to general systems theory (the second wave). It models societies as complex dynamic systems employing multi-agent system technologies. It successfully contests the reductionist, individualist challenge posed by economics by showing the insufficiency of methodological individualism and demonstrating the necessity of realism about macro-level causal powers. Reading this book is a frustrating experience. The author brings together a very diverse collection of debates in the social sciences, but not much constructive emerges from this exercise in name-dropping. This is partly due to the organization of the material. Apart from much repetition, most of the chapters read like advertisements for the new paradigm 1
without revealing much about the actual product. As an example, consider Sawyer s discussion of Durkheim s collective representations in Chapter 6. He dismisses as too individualistic the common way of understanding collective representations as shared mental representations. According to his emergentist interpretation, collective representations are of qualitatively different nature than individual representations because they are emergent social facts (106). And that s all. He does not spell out how the assumed sui generis nature of collective representation could be explicated by using the notion of emergence. Nor does he provide any evidence in support of his interpretation of Durkheim. There are only two places where one gets a glimpse of Sawyer s notion of emergence. One is his discussion of non-reductive individualism, and the other is his discussion of improvisational dialogue. The problem is that these two ideas of emergence are completely different. The latter is based on the idea that the outcomes of improvised dialogue cannot be predicted by (or reduced to) the prior intentions of the participants. The idea that dialogue has unintended effects sounds plausible: one can say that something novel emerges via the interaction. However, I have problems seeing how this non-controversial temporal notion of emergence relates to the strong notion of emergence that Sawyer borrows from the philosophy of mind. Sawyer s non-reductive individualism is based on an analogy to non-reductive materialism in the philosophy of mind. This position accepts ontological individualism, according to which only individuals exist. However, it does not advocate reductionism: higher-level (social) properties cannot be reduced to lower-level (individual) properties. Only the theses of supervenience and token identity hold: every token social event is identical to some individual-level event. The properties at the social level have autonomous causal force. Sawyer s ultimate argument against reductive individualism follows closely Fodor s arguments for the autonomy of intentional psychology. According to this argument social 2
properties are heterogeneously multiple realizable, which prevents the formulation of meaningful bridge-laws between kinds belonging to these two levels. This makes the reductionist program hopeless. More ambitiously, Sawyer also argues that there is downward causation from the social level to the individual level that cannot be accounted for by individual-level laws. In his view, the acceptance of social causation presupposes that one rejects the causal completeness of individual-level causation. More precisely: it could be the case that a social property S, with supervenience base I at time t 1, could lawfully be identified as the cause of social property S* and individual property I* at the time t 2, even though I cannot be lawfully identified as the cause of I*, due to wild disjunction. If the individual-level equivalent of a social property is wildly disjunctive, then explanations in terms of I and I* are not necessarily lawful, even when the relation between S and S* is lawful. (70-71) I find this line of argumentation both troubling and inconclusive. It is troubling for a number of reasons. First, Sawyer provides no argument to support his claim about wild disjunction. He takes it for granted that the bases of social properties are heterogeneous and numerous in a manner that prevents their classification as a manageable set of kinds. Second, the analogy between physicalism and individualism does not work. Physicalism is a comprehensive thesis which states that all non-physical facts are without residue based on physical facts and that the physical realm is causally complete: we do not need to appeal to outside of physical realm to account for physical facts. When we translate this to the case of individualism, we get something like the following: all social facts are without residue based on facts about individual agents, and the realm of individual agents is causally complete, in the sense that we do not need to appeal outside of it to account for facts about individual agents. Both of these claims are evidently false. The causal bases of social facts certainly include facts that are not about individual agents. For example, non-human biological organisms and technological artifacts are certainly constitutive parts of societies: were they different or absent, the social facts would be different. Furthermore, the realm of intentional psychology 3
is not causally complete in the same sense as the physical realm is supposed to be. I thus conclude that the analogy does not really work. There are also problems with the argument itself. If S is token-identical with I and it causes I*, then it follows that also I is a token cause of I*. If supervenience holds, then there could have been some alternative individual-level causes of I*. However, this does not change the fact that there is still a lawful connection between I and I*. Sawyer s argument does not even get started, as there is nothing problematic in the notion of alternative causes, even if one accepts the regularity account of causation as Sawyer does. It is strange that Sawyer wishes to deny the reality of individual-level laws, since it would have been enough to show that social-level causal claims cannot be reduced to individual-level claims. Even if one accepts the existence of a lawful relationship between I and I* it does not directly follow that the supposed lawful relationships between S and S* and S and I* can be reduced to it. Consequently, there might be some room for legitimate social-level causal claims. Rather than challenging the assumption that the existence of lower-level causes excludes the possibility of higher-level causes, Sawyer ends up defending a much stronger argument about downward causation. The basic problem is that these stronger claims about emergent causal powers do not really make sense without giving up the assumption of ontological supervenience, which was the starting point of the argument. The above argument is founded on the idea that there are covering laws to be found at both the individual and social levels. I find it troubling that Sawyer neither discusses what kinds of things these covering laws are, nor provides any examples of them. Of course, his argumentation could be regarded as hypothetical: if there were such laws, emergent social-level causes could be empirically possible. But this kind of modal argumentation provides a very weak basis for Sawyer s research program. The problem is amplified by the fact that his argumentation is tied to very specific assumptions about causation, explanation 4
and reduction. If one gives up the covering-law approach and the Nagelian account of theory reduction for something more scientifically plausible, the whole pattern of argumentation loses its point. Given how comprehensively philosophers of science have shown the shortcomings of both of these accounts, it is surprising that philosophers of mind as well as Sawyer are still attached to these ideas. At the very least, one would have expected that in a book-length study of social emergence, the author would have at least discussed these challenges to his foundational concepts especially because the notion of emergence is defined in terms of these concepts. If one wishes to make a plausible case for emergence, one should start with the deeply contested concepts that are used in its definition. Without clarifying the notions of causation, level, reduction and explanation, there is no hope for dropping the label of conceptual confusion from the notion of emergence. Sawyer s strategy is the complete opposite of this: he does not discuss at all what he means by these notions. Consequently, his key concepts like emergence and downward causation remain opaque and open to alternative interpretations. In this context it is surprising that Sawyer dismisses as trivial William Wimsatt s account of emergence as non-aggregativity. This is very unfortunate as Wimsatt s ideas could have been quite useful for his purposes. Contrary to Sawyer s interpretation, Wimsatt s point is not to provide a new definition for the concept of emergence, but to provide concepts for understanding the different ways in which wholes could be more than their parts. After going through 8 chapters that do not really get to the point, the reader is bound to be disappointed with the final chapter. Rather than providing a substantial account of emergence, it presents an extremely stylized history of recent sociological theory in terms of three paradigms: the structure paradigm, the interaction paradigm, and the emergence paradigm. Obviously, the emergence paradigm comes out as the winner, as the two others 5
have not focused enough attention on social emergence, which according to Sawyer is the central foundational question in the social sciences. Only on pages 210 to 223 does Sawyer provide any explicit discussion of social emergence. His main innovation is the distinction between epheremal and stable emergents. In Sawyer s five-level ontology, these levels are located above the levels of individuals and interaction and below the level of structure. Examples of epheremal emergents are interactional frames and participation structures, whereas group subcultures, conversational routines and collective memory are examples of stable emergents. These two levels continually constrain the flow of interaction and, at the same time, are emergent collective social facts that can be characterized independently of participants interpretations of them. According to Sawyer they have independent ontological status and their own unique causal powers to enable and constrain individual actions. I have a feeling that I understand what Sawyer is trying to say, but the trouble is with his notions of level, emergence and causal power. I fail to see how these notions can shed any light on the problem. Despite its ambitions, Social Emergence fails to be an important contribution to the philosophy of the social sciences. It attempts to bring together many interesting bits of social scientific theory, but the potpourri that emerges is less than a sum of its parts. The argumentation is sketchy, the underlying philosophy of science is outdated, and its central concepts are confused. However, the book shows the power of the notion of emergence: it can create conceptual confusion not only in the philosophy of mind, but also in the philosophy of the social sciences. Petri Ylikoski Department of History and Philosophy University of Tampere Finland petri.ylikoski@uta.fi 6