There are at least two reasons to recommend this book to the readership of the journal. First of all, this is not a reader or textbook on the philosophy of social science, but a collection of ten (mostly) original essays by contemporary philosophers, each commented by a practicing social scientist. The book's format is motivated by Mantzavinos's view that ‘the philosophy of the social sciences cannot ignore the specific scientific practices [. . .] and will only be valuable if it evolves in constant interaction with the theoretical developments in the social sciences’ (p. 1). The book is valuable as a record of such rare interactions between philosophers and social scientists (the book is based on a conference ‘Explanation in the Social Sciences: Philosophical Theory and Scientific Practice’ held at Universität Witten/Herdecke in 2007). Secondly, and more importantly for the readership of this journal, ‘the social sciences’ in the book's title include economics. Economics is often de facto excluded from the ‘philosophy of (the) social science(s)’, and this, as I see it, has to do with the fact that many philosophers of social science have regarded the methodological autonomy of social science vis-à-vis natural science as something they should defend, an agenda that most philosophers of economics do not share. Mantzavinos's aim is to part from this natural-social dichotomy, by pushing further ‘the frontiers of the philosophy of the social sciences as a sub-discipline of the philosophy of science’ (p.1: my emphasis). So the volume is relevant for philosophers of economics, who, in this sense, are at the same time philosophers of social science. In what follows, I will review each chapter in turn, and evaluate (i) how successful the book is in reorienting the identity of the philosophy of the social sciences, and (ii) whether the format of the book is instrumental in achieving this goal.
Part I of the book, consisting of three chapters, concerns ‘Basic Problems of Sociality’, in particular social ontology and collective agency. John Searle, in the first chapter ‘Language and Social Ontology’, argues that language plays a crucial role in the creation, constitution, and maintenance of social reality. Searle's discussion, although instructive in showing a connection between his philosophy of language and social ontology, is nevertheless in many ways unsatisfactory. The main problem is that it is far from clear how Searle's logical or linguistic analysis of the creation, constitution, and maintenance of social reality is related to the causal analyses of the same processes offered by the social scientists. For example, Searle says that ‘the deontologies already present in language are easily and inevitably extended to create such institutional facts as those involving private property, marriage, and power relations’ (p. 18: my emphasis), but that this is not logically necessary. In order to explain why something not logically necessary follows easily and inevitably, Searle resorts to the practical advantages of having those institutional facts. This explanation strikes me as ad hoc in neglecting the fact that language itself is a social institution which has evolved partly because of its practical advantages. Thus it is far from obvious that other social institutions have evolved as a ‘natural extension’ (p. 17) of deontologies created by language. David Lewis's (Reference Lewis1969) game-theoretic analysis of convention, for example, provides an alternative approach which does not assign a crucial theoretical role to language, but instead explains language just as another social institution. The utility of Searle's ‘language-first’ approach might seem limited for those who are familiar with the game-theoretic approaches (evolutionary or not) to the study of social institutions such as property rights and social norms that have been developed by philosophers and economists since Lewis. In his comment, the cognitive scientist Mark Turner praises Searle's refusal to take language for granted, and provides a list of other things that are (naively in his opinion) taken for granted in the social sciences. Turner criticizes, among others, economic theory as a folk theory that fails to explicitly question a stable personal identity ‘with preferences that drive choice toward outcomes’ (p. 36). However, in rejecting economic theory for its apparent inconsistency with the latest cognitive scientific theories, Turner reveals a deep disagreement with Searle, who claims that ‘where the social sciences are concerned, social ontology is prior to methodology and theory’ (p. 9). From this Searlen point of view, Turner's implicit assumption that cognitive science, but not economics or any other sciences, provides the right theoretical apparatus for investigating society needs some defence.
In chapter 2, ‘Shared Agency’, Michael Bratman offers a model in which shared intention among the members of a small group (his example of ‘modest forms of sociality’) is constructed out of the conceptual building blocks provided by his theory of individual planning agents. Again, the relevance of Bratman's discussion to the social sciences is not clear. In his comment, the sociologist Pierre Demeulenaere suggests that characterizing norms of individual rationality is itself a social process that can reinforce particular types of behaviour. To the extent that there is such performativity in norms of rationality, a conceptual reduction of the social to the individual (or a construction of the former from the latter) on its own cannot illuminate the actual interactions of the social and the individual processes in which social scientists are interested.
Philip Pettit's chapter, ‘The Reality of Group Agents’, demonstrates more convincingly the relevance of philosophical theory to the social sciences. Pettit's goal is to show ‘how far groups of individuals can constitute agents on a par with individuals’ (p. 67, fn. 1), and his strategy proceeds two steps: first to propose indisputable criteria for something to count as an agent, and second to demonstrate that some institutionally relevant class of groups (e.g. straw-vote assemblies) satisfy those conditions, thereby counting as genuine group agents. Although very briefly, Pettit also discusses some implications of his results for social theory, e.g. the rejection of explanatory reductionism (to individuals) in the social sciences. Not only is Pettit's analysis ‘a powerful contribution to one of the fundamental problems in the philosophy of the social sciences’ (p. 96), as Diego Rios comments, it is also an important social scientific work that has relevance to both research and practice.
Part II of the book ‘Laws and Explanation in the Social Sciences’ mainly concerns the methodology of the social sciences, and consists of four chapters. David Papineau's chapter ‘Physicalism and the Human Sciences’ tries to summon the ghost of physical reductionism in the social sciences, which Jerry Fodor convincingly laid to rest more than three decades ago. Predictably enough, the commentators Robert Shulman (neuroscientist) and Ian Shapiro (political scientist) dismiss the chapter as ‘a philosopher's game’, arguing that Papineau's reductionism is not only irrelevant but also inimical to contemporary scientific research, which mostly concerns empirical tests of useful macroscopic theories. Although I am sympathetic to their dismissal, a little more internal, philosophical analysis of the argument will be necessary to show what exactly is wrong with it. Papineau formulates a ‘methodological worry’ for the social sciences as the following syllogism (p. 113):
-
Premise (I) The categories of the human sciences are variably realized [at the physical level].
-
Premise (II) Physicalism [which is true] implies that all laws must be reducible to physical laws.
-
Conclusion (III) There are no laws in the human sciences.
Papineau agrees with Fodor on Premise (I), but disagrees on the way in which he argues from the truth of Premise (I) to the falsity of Premise (II). That is, Papineau claims that, contra Fodor, variable realizability does not make reducibility of human laws into physical ones unlikely. The disjunction of the physical realizations of human categories, Papineau argues, forms a physical kind because human categories are systematically related to physical ones (p. 111). However, this argument is problematic in my view. Physicalism states that the causal influences of human categories are exhausted by the causal influences of its physical realization, but this in-principle statement alone does not guarantee that a human category is in fact systematically related to physical one(s), a claim that needs to be empirically established. Perhaps being aware of this, Papineau ventures to make an empirical argument that Gresham's Law, Fodor's example for Premise (I), has in fact a uniform physical realization, because (a) the category of money, although embodied in diverse physical states, can be reduced into a uniform psychological state: ‘X is money if and only if the population in question expects to be able to exchange X for goods and services’ (p. 114); and (b) such psychological state can be explained in terms of ‘the psychological principles of decision theory’ (ibid.) concerning preferences and expectations, which, in turn, may well be uniformly realized at the physical level within humans. Again, this argument is unconvincing because Papineau has no substantial evidence to support premise (b). In fact, a glance at the recent debate on neuroeconomics suggests that neither proponents nor critics of this research programme support the reduction of preferences and expectations to uniform neurological (not to mention physical) states. On the one hand, the proponents are suggesting that, based on new brain-imaging data, we can eliminate or re-write economics constructs such as preferences and expectations, not to reduce them to some co-extensive neurological or physical categories. On the other, the hard-core critics contend that brain-imaging data leave economic theory completely unaffected. Nowhere in the spectrum of this debate can be found a scientist arguing for the kind of type-type reduction that Papineau is suggesting. To sum up, Papineau's criticism of Fodor's argument is unsuccessful both on conceptual and empirical grounds.
Independently from the critique of Fodor, Papineau concedes that after all there are laws in the social sciences (i.e. Conclusion (III) does not follow because Premise (II) is strictly speaking false). However, he tries to reinstate his methodological worry in another way. The alleged worry is that many laws in the social sciences may be ‘selectional laws’, which lack the projectibility of the laws of physics and chemistry. Selectional laws arise as a result of selection processes that favour particular traits over others, and such processes allow for variable physical realization of the same trait (e.g. some insects and birds have been selected for their ability to fly, but their wings are very different from an anatomical point of view). Papineau argues that selectional laws lack projectibility because they ‘tend to stand on their own, rather than fitting into networks of interrelated laws’ (p. 117). However, he does not consider the possibility that a selectional law (of a Darwinian kind, for instance) can explain a wide range of phenomena with a minimum concern about detailed physical implementations. Such a powerful selectional law derives its projectibility from the presumed selective pressures, not from its physical implementations. If this is true then the fact that the law ‘stands on its own’ cannot be a problem for its projectibility.
Sandra Mitchell's chapter, ‘Complexity and Explanation in the Social Sciences’, rejects the standard philosophical conception of laws as the standards for knowledge claims in science, based on the observation that ‘[i]nstead of universal, exceptionless, necessary laws, what one finds in scientific practice [in biology and social sciences in particular] is a range and variety of models, explanations, and theories that provide us with the tools for explaining, predicting, and intervening in our world’ (p. 135). She points out that the standard conception categorizes most of these cases as non-laws partly because it is based on the dichotomous representational framework of standard logic (p. 133), whereas ‘[t]he stability of the conditions upon which a causal relationship depends constitutes a continuum, rather than a singly partitioned space of the necessary and the contingent’ (p. 138). Causal relationships in biological and social domains belong to the more contingent side of the continuum. Mitchell also points out that the sources of contingencies in the biological and social domains are causally heterogeneous, and therefore the detailed understanding of these contingencies are essential if we want to use a regularity in one context for predicting, explaining, and intervening in other contexts. She rejects the ceteris paribus modification of the standard conception of laws as a ‘logical maneuver’ that hides the heterogeneity of causal systems underlying different types of contingencies. What is needed, she concludes, is a conception of laws that can address the methodological and pragmatic questions of ‘how do we detect and describe the causal structures of complex, highly contingent, interactive systems and how do we export that knowledge to other similar systems’ (p. 144). Mitchell's article offers an excellent example of how the philosophy of social science can be informed by the philosophy of other special sciences (in this case biology).
The ontological heterogeneity of the social domain also emerges as a central theme of Daniel Little's article ‘The Heterogeneous Social: New Thinking about the Foundations of the Social Sciences’. Based on a brief survey of the recent theoretical developments in sociology, Little submits an ontological thesis, which he calls ‘methodological localism’. According to methodological localism, the individual is ‘a socially constituted actor, affected by large current social facts such as value systems, social structures, extended social networks’ (p. 163), while these social facts have been largely formed by historical contingencies. Little derives several epistemological and methodological implications from this thesis, such as the rejection of reductive explanations, the centrality of causal mechanisms in explanations, the importance and legitimacy of the eclectic use of different social theories, and the limits of generalizations in the social sciences. The political scientist/legal theorist Jack Knight rightly points out that Little's ontological thesis can be best interpreted as practical and empirical, rather than theoretical and conceptual, and that therefore the thesis is open for empirical investigation.
Nancy Cartwright's chapter ‘What Is This Thing Called “Efficacy”?’ concerns how one can make a sound inference from efficacy, the degree of a causal regularity established in randomized controlled trials (RCTs), to effectiveness, the extent to which such regularity holds in the field. The move from efficacy to effectiveness is a special case of exporting generalizations from one context or domain to the other, which Mitchell highlights as one of the central pragmatic concerns of the special sciences. Cartwright in particular considers social policy interventions (e.g. reducing juvenile crimes, improving pupils’ classroom performance, etc.), and reminds practitioners that RCTs, though a powerful tool for precisely and uncontroversially measuring efficacy, cannot constitute proper evidence without a good theory of how presumed causal processes operate in the intended domain, or at least some clear ideas about the causal effect's endurance, identity and compositionality. Although this is the most pragmatic chapter in Part II, Gerd Gigerenzer suggests it's not pragmatic enough in that it neglects realpolitik (at one point Cartwright explicitly says ‘Forget the politics’, p. 205). He provides intriguing examples of how the evidence from RCTs regarding the effect of breast cancer screening (i.e. that the screening doesn't reduce casualties from breast cancer) has been neglected by policy makers in Germany and the USA, to please the interest groups who wanted the screening introduced. Gigerenzer thus reminds philosophers of the fact that policy interventions based on science themselves are part of social processes that need scientific examination (and intervention).
Unlike Part I and II, which concern ontology and methodology, respectively, Part III of the book has no thematic focus, but instead provides three examples of ‘How Philosophy and the Social Sciences Can Enrich Each Other’. In view of the limited space, I will focus only on the first example, ‘Why Do People Cooperate as Much as they Do?’ by James Woodward (the other two are ‘Situations Against Virtues: The Situationist Attack on Virtue Theory’ by Ernest Sosa, followed by an excellent comment by Steven Lukes, and ‘What Kind of Problem is the Hermeneutic Circle?’ by Mantzavinos, with a substantial philosophical follow-up by David-Hillel Ruben).
Woodward's is the longest chapter in the volume (47 pages), and perhaps the most relevant one for philosophers of economics. The first half of the chapter provides a survey of theory and evidence illustrating the limits of two mainstream accounts of human cooperation in social dilemmas: the ‘self-interest in repeated games’ (SIRG) account and the ‘social preference’ account. Woodward then shows several experimental studies pointing towards an alternative account based on the observation that people are heterogeneous in cooperative behaviour. The majority are conditional cooperators, i.e. are willing to (i) cooperate if sufficiently many others do and (ii) punish non-cooperators even if this is not payoff maximizing. Woodward also takes those studies to suggest that understanding the interaction between different types of people is a key to understanding the mechanisms of cooperation in repeated games. As Werner Güth and Hartmut Kliemt complain in their response, this part of the paper is neither systematic (e.g. Woodward does not cite direct evidence of conditional cooperations: see e.g. Gächter Reference Gächter, Frey and Stutzer2007 for a better survey) nor comprehensive (e.g. Woodward does not mention the ‘team reasoning’ account of cooperation in social dilemmas). However, the chapter's real value is in the second part, where Woodward discusses the role of social norms in coordinating people's behaviour. Woodward agrees with some social preference theorists (reciprocal fairness theorists) that the reciprocal behaviour of conditional cooperators crucially depends on their beliefs about other players’ intentions, but further points out that ‘very specific background information (e.g. about baselines with respect to which kind or hostile intentions can be identified) must be shared among the players’ (p. 240) for these beliefs about intentions to be in equilibrium. Shared social norms and institutions, he argues, can exogenously supply this background information by making a particular set of cooperative equilibria salient. This, however, may raise a methodological worry for those who work in the game-theoretic research programme. According to this programme, a state in which people are conforming to a social norm is nothing but an equilibrium of a game, where a set of players’ beliefs about each other's beliefs and preferences are self-sustaining. One might think that explaining an equilibrium (mutual cooperation) in terms of another (norm conformity) results in a kind of explanatory circularity or regress. This worry is not a mere philosophical fabrication, but a substantial one which may be motivating some game theorists’ strategy to explain the emergence and maintenance of social norms and institutions solely based on norm-independent beliefs and preferences. Woodward rightly responds to this worry by noting that a norm-based explanation is not viciously circular as long as the presupposed conformity is not the same as the explained cooperative behaviour. This implies that one can offer a legitimate historical explanation of a certain cooperative practice referring to the relevant preceding norms and institutions (as an example, Woodward mentions medieval Jewish merchants’ reliance on the commercial law). Moreover, to the extent that the cooperative behaviour in question is sustained by some social norm shared among a group of people, one should expect the influence of this norm to be present in the lab as well (Woodward mentions the variable cooperation rates in the laboratory games played by subjects from small, ‘primitive’ societies). Woodward's second point is particularly original in that it explicitly raises a methodological challenge of how to measure precisely (and control for) the influence of out-of-the-lab norms in laboratory games. So it is somewhat disappointing to see the commentators (an economist and a philosopher) appreciate none of these points.
Overall, this book contributes to reorienting the focus of the philosophy of the social sciences towards the actual theories and practices in the social sciences, thereby yielding some important ontological, methodological and pragmatic insights, many of which are relevant for economics. A final remark on the format of the book: although some comments are not addressing the relevant issues of the main articles, this should be expected as those articles are highly philosophical both in style and in nature. To be fair to social scientists, perhaps someone should edit another book with the reverse format, namely, a collection of cutting-edge social scientists’ essays commented by philosophers. Such a book would be a good test case of whether philosophers of the social sciences can say something philosophically interesting without changing the topic of conversation.