In two groundbreaking papers published around twenty years ago, Mauricio Suárez put forward arguments that helped shape the trajectory of the philosophical literature on scientific representation (Suárez Reference Suárez2003, Reference Suárez2004). Other important contributions over the years have shed light on a number of different theoretical and methodological issues, in some cases showing how philosophical understanding of science can be enhanced through historical analysis of scientific practice and in other cases through consideration of relevant practices outside of science. All of these elements are present in his most recent book, which is a must-read for any philosopher working on scientific representation and modeling, and is also highly recommended for anyone else interested in understanding some of the key issues and controversies in philosophy of science today.
The book starts with a concise but illuminating overview of the history of philosophical debates about scientific representation, enriched by brief first-person accounts of Suárez’s interactions with important contributors to those debates. The plural form here is intentional: one of the upshots from chapter one is that, not only are there different philosophical questions and different philosophical debates surrounding scientific representation, but there are also different histories that can be told about these questions and debates. And, in contrast with an ‘official history’ that focuses on the shift from the syntactic toward the semantic conception of scientific theories, Suárez defends a more pragmatic view, proposing instead that the emergence of recent debates about representation was (and still is) more strongly motivated by interest in understanding actual modeling practices and philosophically making sense of models as they are used in particular scientific contexts. As he emphasizes, the significance of this historical perspective lies in how it reorients philosophical theorizing: ‘it means that a theory of representation must be judged primarily by its ability to answer pragmatic questions about modeling practice, not analytic questions regarding the metaphysics of the relation between our thought and the world’ (p. 4). This pragmatic approach is what the rest of the book articulates and defends.
After an introductory first chapter, the subsequent eight chapters are divided into three parts. Part 1 is titled ‘Modeling’ and it offers, in two chapters, a historical, theoretical and methodological overview of modeling practices in the sciences that provides the basis for the discussions to follow. One highlight of this part of the book is Suárez’s account, in chapter two, of what he calls the ‘modeling attitude’. He proposes that the contemporary reliance on model-based research in science can be traced back to the work of a few influential nineteenth-century scientists working in English- and German-speaking circles. And he argues (compellingly, I think) that what those scientists had in common was not a commitment to any deep philosophical conception of representation, but rather, more simply, a similar outlook on how to do science, how to ask questions, and how to build models to answer them: those scientists developed ‘a style of scientific reasoning, a very general framework or template for applying and carrying out scientific work’ (p. 44), and it is this ‘attitude’ that still characterizes much of science today. Chapter three further illustrates the modeling attitude with more recent examples, while at the same time giving shape to the conceptual framework Suárez relies on in the book. This includes a taxonomy of types of models, and the introduction of important distinctions, such as those between representational ‘sources’ and ‘targets’, and between the ‘means’ and the ‘constituent’ of representation, which are further refined and employed in subsequent chapters.
Part 2 is titled ‘Representation’ and comprises four chapters in which Suárez revisits, revises, and expands upon some of the ideas he is most well known for in the philosophical literature on scientific modeling and representation. The main targets of his criticism are what he calls ‘substantive’ and ‘reductive naturalist’ views of representation, which are views that attempt to define model-target representational relations purely in terms of characteristics of the relata (e.g., similarity, isomorphism). In contrast to these, he adopts a ‘primitivist’ and ‘deflationary’ view, according to which this kind of reduction is not viable: the ‘explanatory power’ of representational relations, he proposes, ‘cannot be further analyzed into a more elementary constituent’ because ‘representation is not a robust property or a relation of sources and targets’; rather, ‘the term representation just picks out some functional dependencies within a particular context of inquiry’ (p. 90). Clarifying these points leads to the positive part of the argument, namely his ‘inferentialism’. The idea is something like the following. Scientific representations can be created in many different ways, and models can bear many different relations to the phenomena they represent, unquestionably including, in many cases, relations of structural correspondence, resemblance, and so on. The point for Suárez, however, is that these characteristics and relations are neither necessary nor sufficient for representation. What is necessary (even if not sufficient) is that the model has the relevant ‘representational force’ and ‘inferential capacities’: very roughly, the model needs to point toward the target according to some shared uses and conventions, and it needs to support inferential reasoning about that target. In many cases, this may well be accomplished by means of similarity or isomorphism, but these particular characteristics and relations are not what makes the model represent the target—on the contrary, many things can have those characteristics and relations, and it is only because the model is supposed to represent a specific target in the first place (i.e., the model is meant to be used as a representation) that those characteristics and relations become important. As he puts it: ‘Nothing represents unless it has the capacity to license some surrogative inferences’, to which he adds ‘however minimal, inappropriate, misleading, or incorrect these inferences might be’ (p. 157). This last clause further reinforces Suárez’s rejection of traditional conceptions of ‘representation’ as a success term. In his view representation does not entail accuracy or truth, and this has important implications for debates about misrepresentation, idealization, and abstraction—a key topic that he also addresses in this part of the book.
The four chapters in Part 2, thus, stick to more familiar terrain relative to Suárez’s well-known contributions to the philosophical literature on scientific representation. In contrast, Part 3, titled ‘Implications’, branches out to consider how the inferentialist outlook relates to, and can be applied in, other philosophical debates. Here Suárez considers how his approach to representation connects to philosophical discussions about artistic representation, with special attention to painting (chapter eight), and he also turns to epistemological issues relating to values, the representation-intervention contrast, and the prospects for realism and instrumentalism (chapter nine). In both chapters from Part 3, the discussions are provocative and suggestive, sometimes raising more questions than providing definitive answers—yet, this is not necessarily a bad thing, especially if the questions we end up with are better than the ones we had at first.
One of those lingering questions for me concerns the relation between the aesthetic and epistemological points that Suárez addresses in the two final chapters of the book. Following along the pragmatic thread he started in the Introduction, in Chapter nine Suárez rejects Hacking’s contrast between representation and intervention while still agreeing with the underlying Deweyan critique to the ‘spectator theory of knowledge’: the resulting view is one in which we make sense of the epistemic achievements of scientists (including what they accomplish through representing) by seeing these achievements as a matter of situated acting and doing, engaging and coping with the world, more so than as detached, contemplative thought about the world. I share the same pragmatist sensibility, and it is for this reason that the discussion about artistic representation in Chapter eight came to seem, in retrospect, to be overly conservative.
The idea that scientific representation can be productively approached through the lens of theories of artistic representation has become popular among philosophers of science in the last couple of decades. And while it was instructive to see how Suárez’s inferentialist view of representation can be applied in this context, the choice to focus on painting seemed potentially counterproductive: after all, paintings are static objects that are typically displayed in galleries for passers-by to contemplate and admire, precisely as spectators. Given Suárez’s own stated goals, it would have been interesting to see how inferentialism in a distinctly non-spectatorial epistemological key could reorient even these ongoing philosophical discussions at the science-art interface. Early on in the book Suárez insightfully points out that ‘philosophers’ understanding of the nature and working of models is often second-hand and more akin to the spectator’s view of a game than the fully engaged know-how of the participants themselves’ (p. 19). The point here is that perhaps this is precisely why philosophers of science have tended to be attracted to art, and especially painting; that is, because thinking about scientific modeling in relation to artistic representation sits especially well with this second-hand, spectatorial point of view typical of philosophy.
In contrast, perhaps a broader conceptualization of ‘art’ (even in a Deweyan spirit) could get us outside of the spectatorial domain and consequently closer to understanding the first-hand experience of scientists in the ‘craft’ of building and using models. An obvious candidate here is the alternative philosophical approach which treats models as tools, instruments, and artifacts. Intuitively, if you’ve only ever seen hammers on display on a shelf but never used one, you might know a lot of facts about hammers and even appreciate what a hammer can represent, but you don’t know what it means to actually use one and all that you can learn through practice with it. Artifactualism and related views of models as tools are diverse, but they by and large coincide in supporting a more engaged rather than spectatorial way of thinking about models and their epistemic significance for scientists. I suspect that inferentialism can make a contribution here too, yet the discussion in Chapter nine, spanning only a couple of pages, was too quick and, it seems, unduly dismissive.
To be clear, these comments are not meant as indictments. If anything, they voice recognition of Suárez’s success in going beyond the core of the arguments in Part 2 and identifying exciting new directions for applying the inferentialist perspective. His engaging style and suggestive ideas leave you wanting more, and that’s a good thing. Inference and Representation: A Study in Modeling Science is a mature statement of a way of thinking about modeling practices that Suárez has been developing for a while in different papers, presented here in a more polished, detailed, and nuanced way, with a level of clarity that comes with time and is especially afforded by the book format. Different readers with different interests and backgrounds will surely find valuable insights throughout the book. Those coming to it for Suárez’s arguments against substantive theories of representation and in favour of deflationary inferentialism will not be disappointed. But the book has a lot more to offer and is a highly recommended read.
