Introduction: The Debates on Scientific Realism

What Is at Stake With Scientific Realism?

When asked what we know about, say, combustion phenomena, or living organisms, it is natural to turn to science (in these cases chemistry or biology) for an answer. As can be read in chemistry textbooks, combustion involves reactions where big molecules break into smaller ones and release energy, for example. One could think that science provides an understanding of these phenomena because scientific theories literally describe what exists in the world, because they unveil the nature of reality. This is roughly what the doctrine of scientific realism says.

On the surface of it, scientific realism looks like nothing but common-sense trust in the capacity of science to give us knowledge and understanding. This attitude can, of course, be qualified (only mature theories are concerned, for example), but a general distrust that would concern all of science could seem misplaced, and indeed, scientific realism is often used in the philosophical literature as a mark of seriousness.

However, as I explain in the introduction of Modal Empiricism, I think that such a surface reading of the debate, which places anti-realism in an uncomfortable position, is inaccurate. An anti-realist does not necessarily put into question the capacity of science to produce knowledge. It rather challenges the realist interpretation of science. Is science really concerned with the deep nature of reality? Or is it concerned with something more practical and mundane? Is there really more to explanations and understanding than the functions they play in our interactions with the world? In sum, I believe that the debate on scientific realism is more accurately framed as a debate over the interpretation of science than as a debate about trust versus scepticism.

My aim, in my book, is to present and defend an anti-realist interpretation of science, Modal Empiricism. Empiricist positions are often characterised as the idea that theories do not aim at being true, but merely aim at “saving the phenomena”. As a first approach, modal empiricism can be understood as the idea that theories aim at saving the possible phenomena. This is, so to speak, a reinforced version of empiricism. However, in light of what has just been said, this position should not be understood as a mere form of scepticism in contrast with realism. I take modal empiricism to be the precise articulation of a pragmatist stance towards science, a stance that puts practice at the centre of interpretational issues.

Before saying more about modal empiricism, let us examine the multiple facets of the debate on scientific realism.

The Components of Scientific Realism

Scientific realism is often described as a commitment to three theses: a metaphysical, a semantic and an epistemic thesis (Psillos 1999).

The metaphysical thesis can be understood as the idea that an external, mind-independent reality exists, and that this reality is structured in a way that is in principle intelligible. Metaphysical realism opposes idealism, according to which reality is mental, as well as Kantian views and some versions of constructivism, according to which the phenomena are somehow constituted by our representational activities. These metaphysical aspects are not be discussed at length in the book.

The semantic thesis of scientific realism has to do with the relationship between our representations and reality. There is, firstly, the idea that theories can be true or false, just like descriptions, but unlike tools, which are good or bad rather than true or false. Secondly, there is the idea that theoretical truth is not “epistemically constrained”: whether a theory is true or false does not depend on the users of the theory, their aims and positions in the world, but only on external reality. So, according to the realist, terms like “observation”, “measurement”, etc. should not appear in the interpretation of the theory as unanalysable primitives, because these terms have to do with the users of the theory rather than with reality itself. The chemical description of combustion phenomena in terms of molecules, for example, would not depend, for its truth, on a particular human perspective on such phenomena.

Semantic realism is sometimes characterised as the idea that theories should be interpreted “at face value” (and not “reinterpreted”), which would imply that realists pay more respect to science than anti-realists. I think that this is a mistake. As I explain later in the book, the notion of “face-value interpretation” is far from clear, and a realist conception of truth is not necessarily the best way of making sense of scientific discourse.

In any case, the fact that scientific realism incorporates a semantic component should clarify why the debate on scientific realism can be understood as a debate over how to interpret science. For example, an anti-realist can accept that science gives us knowledge in the form of true theories, but at the same time adopt an epistemically constrained theory of truth that makes her position anti-realist.

Finally, the epistemic thesis of scientific realism can be phrased as firstly the idea that science aims at producing true theories, and secondly the idea that science is successful in its aim. The first idea is focused on scientific activity as such, the question being what is the aim of this activity, or what are its norms. According to the realist, science aims at truth, so the norms of scientific activity, those by which theories and hypotheses are accepted, must be directed towards truth. In this respect, the realist thesis contrasts with Van Fraassen (1980)’s view that science only aims at “saving the observable phenomena”, but also with pluralist approaches, according to which it does not make sense to attribute a unified aim to science. The second question is whether science is successful, and this question has more to do with the end-products of scientific activity, notably scientific theories, than with the activity itself. Whether scientific theories are true or at least approximately true, or whether they actually “save the phenomena”, is related to traditional issues of justification.

Here, we can see a second reason why anti-realism should not necessarily be considered an attitude of defiance towards sciences: if the aim of science is not truth, then one can be an anti-realist and at the same time consider that science is a successful endeavour.

Modal Empiricism: an Interpretation of Science

Why challenge scientific realism? The main reason is, I think, that there is an inherent tension in the position, which is introduced by the assumption that theoretical truth is not epistemically constrained. It implies a principled gap between truth and our capacities to know that something is true.

The tension is particularly salient in the case of science, because arguably, systematic confrontation with experience is the cornerstone of scientific methodology. But how can we claim to have knowledge of universal laws of nature if those are only ever confirmed by limited sets of observations? And how can we claim to have knowledge of unobservable entities such as electrons if their existence is only ever confirmed by their indirect consequences on our observations? What is at stake is the validity of ampliative modes of inference, that is, inferences that go beyond mere appearances.

All this could seem like unfounded philosophical worries given the astonishing empirical successes of science. However, being willing to abandon theories, even very successful ones, in the face of new evidence is at the core of scientific methodology, so why should we be so confident that electrons and gravity will never be replaced by different concepts? Success versus theory change: this is, in a nutshell, how the contemporary debate on realism is framed.

I believe that modal empiricism fares better than its competitors with respect to these two arguments. The position can be summarised as follows:

Modal empiricism (1)

the aim of science is to produce theories that correctly account, in a unified way, for all our possible manipulations and observations within particular domains of experience, and science is generally successful in this aim.

This position challenges the epistemic thesis of scientific realism presented above. However, it is still committed to the idea that science has a unified aim, which is modal empirical adequacy. In this sense, it is close to other versions of empiricism.

The first main difference between modal empiricism and other versions of empiricism lies in its reference to possible phenomena, whereas other versions typically focus on regularities in actual phenomena. This aspect should be understood in terms of natural modalities, that is, not in terms of what is or is not conceivable or likely to be the case, but in terms of what is or is not possible in this world given natural constraints on phenomena. The second difference is the reference to manipulations, and not only to observations, and I will explain how these two differences are intimately related. In sum, according to a modal empiricist, an adequate theory tells us what it is naturally possible to do and observe in this world.

The notion of possibility that characterises modal empiricism is quite distinct from traditional construals of natural modalities, for example in terms of absolute laws of nature, because the possibilities we are talking about are, so to speak, anchored to actual contexts. I will propose in the book an understanding of these situated modalities in terms of possible situations. I will argue that this understanding is what makes the position distinctively anti-realist, since the contexts to which we have access are limited by our situation in the universe and by our cognitive constitution, and that it is also what allows modal empiricism to be the best compromise in the debate on scientific realism, because it allows to explain the success of science without relying on problematic modes of inference.

In the conclusion of the book, I suggest a possible reformulation of the position as challenging the semantic thesis of scientific realism instead of the epistemic one. Adopting a different, pragmatist notion of theoretical truth, which happens to coincide with the modal version of empirical adequacy just presented, it is possible to maintain that indeed, science aims at truth and that it is successful in this respect. However, this is not the notion of truth that the scientific realist has in mind. This reformulation is a way of making sense of the natural attitude described in the opening of this introduction without being a realist. This gives us another understanding of modal empiricism:

Modal empiricism (2)

our best scientific theories are true, or approximately true, in the sense that they correctly account, in a unified way, for all our possible manipulations and observations within their domains.

As its formulation in terms of pragmatic truth makes clear, modal empiricism is a form of pragmatism. I understand pragmatism as being characterised by an emphasis on a practical, active conception of knowledge, as opposed to what Dewey calls the “spectator conception of knowledge”. The main idea is that in order to interpret scientific theories, we should first have a look at how they are used in particular contexts, and an interpretation should have practical implications in one way or another. I argue in the conclusion of the book that this semantics, captured by our second definition of modal empiricism, is in line with Peirce’s pragmatism, and that it is also consistent with scientific discourse. I also argue that this approach has the capacity to connect metaphysical debates to more tangible considerations.

The Two Battles to Be Fought

As a middle position between scientific realism and traditional versions of empiricism, modal empiricism must be defended against two main opponents: the more pessimistic ones and the more optimistic ones. There are two battles to be fought.

In Modal Empiricism, I first prepare the battle field by reviewing the debates on the nature of scientific theories and models and their relations to users, experience and the world. Then I propose an account of scientific representation that distinguishes the contextual, situated uses of scientific theories and their communal, normative status. This account of representation constitutes the framework on which modal empiricism is based.

I develop, on this basis, the conception of empirical adequacy that modal empiricism endorses, with its characteristic modal and situated aspects. This is a conception of the norms of ideal success by which theories are accepted or rejected in science. I argue that modal empiricism better accounts for actual scientific practice as a rational activity than weaker versions of empiricism. Then I argue that this notion of empirical adequacy can be justified by induction on experience, thanks to the situated aspects of modalities. This eliminates the main reluctance an empiricist could have in accepting that there are possibilities in the world. So much for the first battle.

The second battle is against realists. The main realist strategy consists in inferring theoretical truth from the “miraculous” empirical success of theories, in particular when it comes to novel, unexpected predictions. This strategy involves a particular mode of inference: an inference to the best explanation. I criticise this strategy and propose an inductivist alternative, which rests on a particular form of induction: an induction on models. If empirical adequacy is justified by induction, then novel predictions are no miracle, and this does not imply that theories are true in the realist sense. Finally, I explain how modal empiricism can respond to problems of theory change better than realist positions, including structural realism. This clarifies how modal empiricism differs from realism in its various guises.

The conclusion of the book is an opportunity to return to the facets of the debate that have been left out in other chapters: the semantic and the metaphysical aspects. Semantic realism is rarely discussed in the literature, but I think that a pragmatist alternative, based on modal empirical adequacy, can make better sense of scientific discourse. This final move has important implications for the metaphysics of science. It favours adopting a revisionary pragmatist stance towards metaphysics, by connecting metaphysical considerations to more tangible aspects, without falling back to an impoverished metaphysics.

Modal empiricism should not be merely understood as yet-another-position-in-the-debate-on-scientific-realism, but it should be understood as a more far-reaching proposal concerning the way of interpreting science in general. This proposal has a normative dimension with regards to the metaphysics of science: sound metaphysical speculations should seek systematic connections with possible experiences in order to have practical relevance. In sum, modal empiricism purports to be the precise articulation of a pragmatist approach towards the philosophy of science. I hope that it can serve as a useful basis for all philosophers sharing this pragmatist stance.