10 REDUCTION THAT REALLY MATTERS Can We Reduce from Macro to Micro in the Social Sciences (And Should We)? José A. Noguera Affiliation needed. P 1. Methodological Individualism, Reduction, and Analytical Sociology rejudices on what others think are usually widespread in society, and also among scholars in the social sciences. For instance, nonanalytical sociologists tend to think that all analytical sociologists are methodological individualists and advocates of rational choice theory. They would be astonished to see how much criticism against those options is launched in analytical sociology conferences (as I was when I first came into the analytical sociology circles). It could in fact be claimed that the harder and deeper critiques to methodological individualism1 have come from those who are supposedly their most dogmatic defenders in the eyes of the rest of the sociological community (this is hardly realized, for example, among Bourdieu s followers). Surprising as that may be, the fact is that the inspiring fathers of analytical sociology such as Boudon, Elster, or Coleman have very 1
epistemol o gy of the so cial sciences clearly subscribed to one or another version of methodological individualism (MI henceforth). A well-known and useful definition of MI is the one given by Elster: explanations in the social sciences should refer only to individuals and their actions (2007: 13); in an earlier work, he proposed a more detailed definition which reads like this: By methodological individualism I mean the doctrine that all social phenomena (their structure and their change) are in principle explicable only in terms of individuals their properties, goals, and beliefs. This doctrine is not incompatible with any of the following true statements. (a) Individuals often have goals that involve the welfare of other individuals. (b) They often have beliefs about supra-individual entities that are not reducible to beliefs about individuals. The capitalists fear the working class cannot be reduced to the feelings of capitalists concerning individual workers. By contrast, The capitalists profit is threatened by the working class can be reduced to a complex statement about the consequences of the actions taken by individual workers. (c) Many properties of individuals, such as powerful, are irreducibly relational, so that accurate description of one individual may require reference to other individuals. (Elster, 1982: 453) So defined, I always thought MI was a trivial truth, as Elster himself has often noted. Why has it been, then, the object of so much harsh criticism and disdain? One of the accusations most frequently thrown against MI is that it is a reductionist strategy. I think this is right (Elster does too: see 2007: 13, 257 259): reduction consists on explaining macrolevel phenomena in terms of micro-level, more elemental phenomena, so to explain macro-social phenomena in terms of individual behaviour is a reductionist strategy. However, I confess I share the naive belief that scientists are in the business of reduction (among others) and that if science is worth the effort, then reduction is a good thing. I share Elster s statement that reductionism is the engine of progress in science (2007: 258). It is true that the words reduction and individualism have been given so many different meanings and interpretations that disputes 2
reduction that really mat ters around them are highly poisoned by deep misunderstandings. It is probably not very intelligent to still try to save them in no-matter-how nuanced or deflationary ways. However, I do not have better terms to express an idea which, to my view, is sound and trivially true; so, since I am no friend of changing conceptual labels in science for marketing reasons, I will rather stick to those terms here. That MI is a discredited label may be shown by the fact that even analytical sociologists have felt the need of differentiating themselves from its supposedly crude version in Elster s work, by saying they are rather structural individualists ; however, one might have trouble to see what is the conceptual difference between Elster s MI and the definition of structural individualism as formulated in The Oxford Handbook of Analytical Sociology (2009): structural individualism is a methodological doctrine according to which all social facts, their structure and change, are in principle explicable in terms of individuals, their properties, actions, and relations to one another (Hedström and Bearman 2009: 8). To this point, the definition seems clearly similar to the one previously quoted by Elster; to say the least, there is no incompatibility whatsoever between both definitions. But the authors immediately add: It differs from traditional notions of methodological individualism (e.g. Elster 1982) by emphasizing the explanatory importance of relations and relational structures (Hedström and Bearman 2009: 8). And, only a page later, they state: That relations are important for explaining outcomes does not mean that they are independent of individuals and their actions, however. As emphasized above, in principle all relational structures are explainable as intended or unintended outcomes of individuals action. 3
epistemol o gy of the so cial sciences Again, it is clear that no logical incompatibility between this emphasis on relations and Elster s definition of MI exists: point (c) of the above quote from 1982 states without doubt that in Elster s view relations among individuals and relational properties (and, therefore, relational structures ) are legitimate candidates to be included in the explanandum by social scientists. Perhaps realizing this, Hedström and Bearman present the difference just as a stronger emphasis on relations. It may be surprising, though, that a question of emphasis, which is thus implicitly denied to have a conceptual or logical status, justifies the adoption of a new conceptual label. However, as suggested above, I realize that it may be a smart move of frame in the context of the present discredit and widespread prejudice against MI. Reductionism has also been largely criticized and is often used as a pejorative label which dispenses from further arguments against an author or position.2 Within the analytical circles, this feeling is also growing. Take for example the recent and insightful contributions by Daniel Little (2012) and Gianluca Manzo (2012) on analytical sociology: Little defends a so-called methodological localism as an alternative to classical MI and to structural individualism, and one of the motivations to do that is to avoid reductionism (Little, 2012: 149). Manzo brilliantly argues that methodological localism is similar in any relevant sense to structural individualism, and he aims to refute the charge of reductionism on analytical sociology. Both, therefore, seem to depart from the assumption that reductionism, whatever it is, is a bad thing. So did Mohamed Cherkaoui in Chapter 5 of his Invisible Codes (2005), entitled The Multiple Levels of Reality, and subtitled Is the macro always based on the micro?. There, the following statement can be found: Social actionists rediscovered the idea dear to Weber and Popper that all social phenomena (macroscopic level), particularly the structure and function of institutions, should be explained and understood as the result of individual actions, not in terms of universals or collective properties. But in affirming this epistemological principle, actionist theorists did not reduce social phenomena to psychological ones. 4
reduction that really mat ters Cherkaoui was depicting here a very usual view among sociologists in the analytical tradition, and even beyond: a sort of broad antireductionist individualism, allegedly not to be mixed up with strict MI. In what follows, I will schematically argue two points: first, it is not especially problematic in sociology to endorse reductionism in a very simple (and maybe trivial) but important sense: ontological reduction. Second, contrary to the prevalent view, ontological and methodological individualism may have little sense as independent positions. 2. What Does Reduction Mean? Reduction as a Scientific Virtue Reduction is, as suggested above, a term that has been given many different meanings. Of course, as remarked by Elster, different versions of reductionism (or any other research strategy) may be bad and good, and there is no point in stating that bad reductionism is a bad thing (Elster 2007: 259). But most critics seem to deny the very possibility of any good reductionism in social science. Whether they are individualist (under any of the existing versions) or not, most social scientists seem to agree that reduction is an unfeasible and/or dangerous aim. However, one may argue that reductionism should be a good thing for science. The reductionist stance leads to study the world as a set of assembled parts or elements that may be disassembled (in a material sense or using mental abstraction) and studied separately. A reductionist explanation of a given phenomenon tries to clearly identify its constitutive parts and to analyse how the properties of these parts and the interaction between them have produced the phenomenon. Reductionism, so understood, is one of the fundamental pillars of modern science, as Wilson nicely reminds us: The cutting edge of science is reductionism, the breaking apart of nature into its natural constituents. The very word, it is true, has a sterile and invasive ring, like scalpel or catheter. Critics of science sometimes portray reductionism as an obsessional disorder, declining toward a terminal stage one writer recently dubbed reductive megalomania. That characterization is an 5
epistemol o gy of the so cial sciences actionable misdiagnosis. Practicing scientists, whose business is to make verifiable discoveries, view reductionism in an entirely different way: It is the search strategy employed to find points of entry into otherwise impenetrably complex systems. Complexity is what interests scientists in the end, not simplicity. Reductionism is the way to understand it. The love of complexity without reductionism makes art; the love of complexity with reductionism makes science (Wilson 1998: 58 59). Notice that this is not equating reduction and explanation. Reduction shows that some entities are composed of other more elementary entities, and that this fact is relevant for explaining how they are produced and why they exist. Here is relevant for is important: the reduction is not the explanation by itself, but it helps to explain. You have not explained a mob when you show that its reduction base are individual actions or mental states; you did it when you showed which exact composition of actions and mental states, causally connected in an specific way, gave birth to the mob behaviour, as for example in Granovetter s (1978) threshold models of collective behaviour. If you know the reduction base, you still have to work on how to explain the phenomenon departing from that base; but if you do not know the reduction base, your explanation will be incomplete or false. Reductionism is a scientific virtue because it improves intelligibility in scientific explanations. To reject reduction as an aim in a scientific discipline is to accept that there are ontological gaps in reality, that some phenomena may be self-caused or constituted out of nothing, as if they were separated from the network of causal and/or constitutive relationships that connects all phenomena. To use Weinberg s expression (1992: 56), when a particular reduction is achieved in science, we are a little bit closer to the point where all our explanations will ideally converge. Our explanations have arrows that point to places where we can keep on asking why ; to reduce is to successfully go down this chain of whys, by making disappear the gaps in it. Is this also true for the social sciences? At first sight, one would say that if they made substantial progress along reductionist lines, this could only be considered as a progress in their scientific maturity. However, 6
reduction that really mat ters many social scientists since Durkheim have feared that any reductionist achievement would be a step towards the elimination of their disciplines. But is reduction necessarily accompanied by elimination? It does not seem so.3 We may just observe that if many sociologists fear or reductionism was justified, chemists or biologists would have lost their jobs years ago to give way to physicists. Although reduction ideally aims to the convergence of all explanatory arrows into a common source, that ideal does not mean that all scientific disciplines except one must disappear, because, among other reasons, many of these arrows may depend on knowledge provided by different disciplines. The most reasonable and reduction-friendly model for the integration of scientific disciplines is not elimination but consilience (Wilson, 1998): the interconnection between events and theories based on facts studied by different disciplines, so that a common explanatory ground is generated. Concepts of Reduction However, the answer to the question on elimination depends on the meaning we ascribe to the concept of reduction. It is true that reduction in the sense described so forth (the Elster Wilson Weinberg sense) does not entail elimination. But, among many others, Udehn claims in an influential book that scientific reduction means elimination without remainder (2001: 324 325). The point is he is using a different concept of reduction. So what may reduction mean? Which version of reductionism is most suitable for the social sciences? Which are the reducibility criteria that a reductionist social science should satisfy? It is crucial to answer this question, since, as noticed by Kim (1999: 13), most of the objections against reductionism assume empty or inadequate models of what reduction entails in real scientific practice. Here I will broadly rely on Searle (1992: 122 126) in order to distinguish between three types of reduction: ontological, theoretical, and definitional.4 a) Ontological reduction. What I will call, for simplicity, ontological reduction is a mix of two claims. First, an ontological one: objects of certain types are nothing but configurations of objects of other types. Second, a causal or methodological one: the causal powers of a given entity are explainable in terms of the causal powers of other entities 7
epistemol o gy of the so cial sciences which conform it. I realize that it is highly polemical to merge both kind of thesis into a single concept of reduction, since it is very usual to agree with ontological reduction but not with methodological one (or the reverse). However, I have problems to accept this separation between ontology and methodology, so I will deal with this specific issue in the next section. Here it will suffice to say that, according to Searle, the most important form of reduction is ontological reduction, since it is the form of reduction to which the other forms point to (1992: 123). In a similar way, Moulines has recently remarked that the concepts of reduction and emergence clearly appear to have primarily an ontological meaning (2006: 314). I think this is also the case for the social sciences. The fact that ontological reduction is the most interesting form of reduction from a scientific point of view is fully compatible with the fact that it may be philosophically trivial, because it does not pose important philosophical problems. Instead, theoretical and definitional forms of reduction are the ones that usually produce philosophical problems. b) Theoretical reduction. We have a theoretical reduction when all laws and propositions in a theory or discipline are derivable from the laws and propositions of other more fundamental theory or discipline; in such a case scientists often say that the first theory or discipline has been reduced to the second, or simply eliminated. However, this is a very unusual case in science (Searle, 1992: 123 124). In a classical article, Philip Anderson (1972), showed the difficulty of generalizing this requirement to all scientific theories, because the aspiration to derive all laws and scientific facts in the world from a simple set of ultimate principles or equations generates enormous computability problems given a certain level of complexity. The famous Laplace s demon, hypothetically able to undertake such a project, is nothing more than a mental experiment. One way of avoiding usual misunderstandings is therefore to distinguish the requirement of theoretical reduction (which, to use Anderson s term, could be regarded as a sort of logical constructionism rather than reductionism) from ontological reduction. This distinction helps to see that systematic non-derivability of scientific propositions at the macro-social level from those at the micro level in the social sciences, as well as in many others, does not entail the impossibility to 8
reduction that really mat ters ontologically reduce the latter to the former; and this second reduction is the one that really matters in science. c) Definitional (or linguistic) reduction. While ontological reduction has to do with the nature of the phenomena under study (and therefore with their proper description and explanation), and theoretical reduction has to do with the logical possibility of deriving some propositions from others, definitional reduction is about the possibility of translation between two different languages or vocabularies. Definitional reduction, according to Searle (1992: 124), is unfashionable, since it only cares about the relations between words or sentences, and how those referring to some kind of entities may (or not) be translated without information loss to others referring to different kinds of entities. There are many cases where this is possible. Trivially, when definitional reduction is possible, then ontological reduction should also be possible. But the reverse is not true: the possibility of ontological reduction (the one which really matters) does not necessarily imply the possibility of definitional reduction (nor, even less, the need for it); only by going back to an atomistic conception of language such as the one shared by logical positivists could we support that implication. The impossibility of definitional reduction in a given case cannot then be a valid argument for the impossibility of ontological reduction. It is however very usual in the social sciences to share an anti-reductionist stance based on the claim that in order to adequately explain their objects social scientists need concepts which refer to macro-social phenomena and which would be impossible, hard, or inconvenient to depict in terms of their microfoundations (such as systemic properties, institutional structures, cultural patterns, and the like).5 To my view, this approach fails to make the distinction between definitional and ontological reduction, and can only appear to be anti-reductionist because it focuses only on the definitional version of reduction. But Weber already noticed that the sociological interpretation of actions is frequently forced to work with such concepts [that is, which supraindividual concepts such as state, firm, organization, bureaucracy, and the like, JAN] ( ) with the aim of achieving an intelligible terminology (Weber, 1922: 12); besides, Weber argued, though it would be 9
epistemol o gy of the so cial sciences possible to eliminate these concepts from usual language by introducing long-winded and wordy descriptions of their microfoundations, we still would face the fact that individual actors think of them as real entities, and that thought may have causal effects on their conduct. This is why Elster s definition of MI soundly recalls that trying to reduce individuals beliefs in supraindividual entities is a category mistake. As any other ontological reduction, MI operates in extensional contexts, not in intensional ones (Elster, 1985: 6). The contrary would force us to treat as real any supraindividual or collective entity social actors may believe in, simply because such belief may have a causal effect on their behaviour. As wisely observed by Dore (1961: 76), claiming that social scientists should take their concepts directly from their data (such as individuals beliefs) would be as absurd as forcing the carpenter to use only wooden saws. The truth conditions of the social scientists explanatory propositions which include agents beliefs are different from the truth conditions of those beliefs. John believes that a collective conscience led him to do action A is true or false independently of the truth or falseness of A collective conscience led John to do action A.6 It seems to me, then, that social scientists could agree without important problems that ontological reduction is a good scientific practice; that it is the form of reduction that really matters in science, and that MI, understood in this sense, is the most adequate form that a reductionist strategy may take in the social sciences. 3. Ontology vs. Methodology: A Dubious Disconnection A seemingly conclusion of the preceding section is that the real discussion about MI and reductionism is an ontological one. Why, then, to speak about methodological individualism? Although the question is legitimate, the issue is purely terminological, and does not affect the core discussion of whether the research strategy depicted in Elster s definition is acceptable or not, irrespective of how we call it. We could change the label to ontological individualism and this, I think, would change little about the validity of the claims made. 10
reduction that really mat ters But my aim in this section is to go beyond terminological discussions. In the previous section I defined ontological reduction as a merge of causal-methodological and ontological claims. However, a very widespread view tends to sharply separate both levels. I have always been puzzled to see how many philosophers and social scientists since Durkheim explicitly or implicitly claim to be ontological individualists but not methodological ones (see as a good example, Little, 2012). The reverse is also true: how come that so many social scientists have been methodological individualists (or reductionists) but not ontological ones? (think for example on many neoclassical economists who implicitly use MI while avoiding any ontological commitment, in a typical Friedman fashion). Here is the question I want to consider: once a stand has been taken for methodological (ontological) individualism, should not one go all the way through and embrace also ontological (methodological) individualism? I tend to think the answer is yes in both directions. The reason is that the price to pay otherwise is unrealism, and that is a high price indeed for those social scientists who at the same time criticize rational choice theory for that reason. Suppose that you claim that ontological individualism is sound ( of course social reality is made up from individual actions ), but that the explanatory strategy of methodological individualism is not. Suppose that, at the same time, you favour the typical analytical sociology objections against rational choice theory, and accuse that theory of lack of realism and indifference to the real mechanisms in place behind individuals behaviour. There is some inconsistency in giving priority to methodology in order to criticize MI, but to ontology in order to criticize rational choice theory. You simply cannot have both at the same time, at least without making your point further. If a social scientist thinks that for methodological reasons it is sound to build explanations that ignore his ontological beliefs about how society is made up of individual actions, then he (more than other social scientists) should be sensitive to Friedman s (1966) defence of unrealistic assumptions in economic theory. He may not like the particular unrealistic assumptions Friedman (or any other individualist) make, but he simply cannot criticize those assumptions for their being unrealistic, since he is an 11
epistemol o gy of the so cial sciences ontological individualist but not a methodological one, so he implicitly admits that the ontological components of social reality (individuals) are of little explanatory interest for him. Suppose now that you are, like Friedman, a methodological individualist, but not an ontological one ( modelling economic phenomena as if they were caused by individual decisions is sound even when I am making no ontological claim that individuals really decide anything, or even exist ). Then realism is also sacrificed, and the door is open for any methodological fiction that approximately predicts some pattern or helps our models to fit the data. In the first case (OI but not MI) one would be saying: I know that social reality is like this but I prefer to represent it like that. In the second case (MI but not OI), one would be saying: I prefer to represent social reality like this, whether it is like this or not (which I do not know or care). I may be wrong, but I have always had a hard time trying to understand why someone committed to scientific realism might ever be sympathetic to any of those positions. To say it differently, it is hard to consider as totally independent things the methodological units of our scientific analysis, and the really existing entities in the world. Would it not be more sensible to assume that, if one supports scientific realism, his methodological and ontological commitments should go hand-inhand, also in the case of individualism? 4. Concluding remarks: who s afraid of reductionism? Let us go back to Cherkaoui s remark quoted in section one: classical action theorists like Weber were individualists (they thought that all social phenomena ( ) should be explained and understood as the result of individual actions ), but they did not reduce social phenomena to psychological ones. How should we understand this sentence? I cannot dare to know what Cherkaoui exactly meant by reduce, but I would bet it was not ontological reduction as it has been defined above, since that would be inconsistent with the mandate of explaining and understanding social reality in terms of individual actions. So what does reduction mean here? My guess is that it may mean theoretical reduction 12
reduction that really mat ters (sociology was not to be eliminated by psychology according to those classic theorists) or definitional reduction (the vocabulary of socialsupraindividual entities would not be replaced by a collection of wordy descriptions in psychological terms). If, on the contrary, Cherkaoui was ascribing methodological individualism to Weber, but not ontological one, as well as supporting that option himself, then they would face the problem I dealt with in section three about the implausibility of a sharp separation between the ontological and the methodological levels, at least for scientific realists like them. Be that as it may, and leaving aside many other relevant aspects of the debate on reductionism and MI, contemporary social science seems in his way to achieve some success on the lines of reduction to lower-level phenomena. This growing success could have the side-effect of reducing epistemological and methodological fragmentation, and enhancing the production of social-scientific knowledge in a controlled and accumulative fashion, something that has been more the exception than the rule in sociology. I have tried to present some thoughts on why sociologists (and, in general, social scientists) should be more sympathetic to reductionism and MI in the defined sense. To be sure, it will be difficult to make disappear all misunderstandings, resistance and prejudices against this research strategy. But I do think it is still worth trying. Notes The Spanish National Plan for R&D has supported this work through the CONSOLIDER-INGENIO 2010 Programme of the Ministry of Science and Innovation (MICINN, Grant No. CSD2010-00034-SimulPast) and through Grant No. CSO201231401 of the Ministry of Economy and Competitiveness. Previous extended versions were presented at the Philosophy of Science Seminar of the University of Helsinki (April 2009) and at the IX Spanish Congress of Sociology (Barcelona, September 2007). I am grateful to the attendants to those sessions for their comments. The text also benefited from comments by Sandra González-Bailón, Ernest Weikert, and Luis M. Miller, in several discussions in the Analytical Social Theory Seminar in Barcelona. 1. I leave aside here the case of rational choice theory, which would deserve another different chapter. 2. For instance, the 5th edition of Ritzer s (2000) widely used handbook of sociological theory, the terms reductionism and reductionist appear six times, 13
epistemol o gy of the so cial sciences 3. 4. 5. 6. always associated to pejorative qualifications or negative assessments of several theoretical orientations, without need to argue why reduction is bad. See Brodbeck (1958) for a defense of this claim for the social sciences, Weinberg (1992 and 2001) for the natural sciences, and Kim (1999) or Jones (2004) for psychology and the philosophy of mind. In fact, Searle identifies five different types of reduction: ontological, ontological properties reduction, theoretical, definitional, and causal. However, the ontological properties reduction, as claimed by Searle himself, is just a sub-case of ontological reduction. Here I will also deal jointly with ontological and causal reduction for reasons explained in the next section. Note also that theoretical reduction corresponds to what Udehn (2001: 324 325) calls scientific reduction, while definitional or linguistic reduction is equal to his philosophical reduction. See for instance the well-known contributions to the debates on MI by Berger and Offe (1982), or Levine, Sober and Wright (1987) In more technical terms, intensionality is defined in the philosophy of mind and language as a property of certain sentences that do not satisfy the conditions of substitutability of coreferring expressions salva veritate and existential generalization (Searle, 2001: 59). Typically, sentences about beliefs, desires and volitions, which are present in many social-scientific explanations of behavior, do not satisfy those conditions, or, to say it differently, are opaque in terms of external reference. References Anderson, Philip (1972) More is different, Science, 177: 393 396. Berger, Johannes and Claus Offe (1982) Functionalism vs. rational choice? Some questions concerning the rationality of choosing one or the other, Theory and Society, 11(2): 521 526. Brodbeck, May (1958) Methodological individualisms: Definition and reduction, Philosophy of Science, 25(1): 1 22. Cherkaoui, Mohamed (2005) Invisible Codes: Essays on Generative Mechanisms. Oxford: The Bardwell Press. Dore, R. P. (1961) Function and cause, in The Philosophy of Social Explanation, edited by Alan Ryan. Oxford: Oxford University Press, pp. 65 81. Elster, Jon (1982) Marxism, functionalism, and game theory, Theory and Society 11(2): 453 482. (1985) Making Sense of Marx. Cambridge: Cambridge University Press. (2007) Explaining Social Behavior. New York: Cambridge University Press. Friedman, M. (1966) The methodology of positive economics, In Essays in Positive Economics. Chicago: University of Chicago Press. Granovetter, Mark (1978) Threshold Models of Collective Behavior, American Journal of Sociology, 83(6): 1420 1443. Hedström, Peter, & Bearman, Peter (2009) What is analytical sociology all about? An introductory essay, in P. Hedström & P. Bearman (eds), The Oxford Handbook of Analytical Sociology. Oxford: Oxford University Press. 14
reduction that really mat ters Jones, Todd (2004) Special Sciences: still a flawed argument after all these years, Cognitive Science 28(3): 409 432. Kim, Jaegwon (1999) Making sense of emergence, Philosophical Studies 95(1 2): 3 39. Levine, Andrew; Sober, Elliott, and Wright, Erik O. (1987) Marxism and methodological individualism, New Left Review, 162: 67 84. Little, Daniel (2012) Explanatory autonomy and Coleman s boat, Theoria, 74: 137 151. Manzo, Gianluca (2012) Full and sketched micro-foundations: The old resurgence of a dubious distinction, Sociologica, 1. Moulines, Ulises C. (2006) Ontology, reduction, emergence: A general frame, Synthèse 151(3): 313 323. Ritzer, George (2000) Modern Sociological Theory (5th edn.). New York: McGraw-Hill. Searle, John R. (2001) Rationality in Action. Cambridge (Mass.) The MIT Press. (1992) The Rediscovery of the Mind. Cambridge (Mass.): The MIT Press. Weber, Max (1922) Economía y sociedad. Esbozo de sociología comprensiva (2 vol.) [Economy and Society: An Outline of Interpretive Sociology], México, Fondo de Cultura Económica, 1964. Weinberg, Steven (1992) Two cheers for reductionism, in Dreams of a Final Theory. New York: Vintage Books. (2001) Facing-Up: Science and Its Cultural Adversaries. Harvard: Harvard University Press. Wilson, Edward O. (1998) Consilience. The Unity of Knowledge, New York, Vintage Books. 15