Theory and Reality

by Peter Godfrey-Smith

  12.2 Approaching Scientific Realism

  The position defended in this book is a version of scientific realism. A scientific realist thinks it does make sense to say that science aims at describing the real structure of the world we live in. Does the scientific realist think that science succeeds in this aim? That is a more complicated issue.

  Formulating scientific realism in a precise way will take a while. And the best way to start is to ignore science for the moment and look first for a more general description of "realist" attitudes.

  The term "realism" gets used in a huge variety of ways in philosophy; this is a term to be very cautious about. One tradition of dispute has to do with what our basic attitude should be toward the world that we seem to inhabit. The simple, commonsense view is that the world is out there around us, existing regardless of what we think about it. But this simple idea has been challenged over and over again. One line of argument holds that we could never know anything about a world of that kind. This debate has carried over into the philosophy of science.

  How might we give a more precise formulation of the "commonsense" realist position? The usual starting point is the idea that reality is "independent" of thought and language (Devitt 1997). This idea is on the right track, but it has to be understood carefully. People's thoughts and words are, of course, real parts of the world, not extra things floating somehow above it. And thought and language have a crucial causal role in the world. One of the main reasons for thinking, talking, and theorizing is to work out how to affect and transform things around us. Every bridge or light bulb is an example of this phenomenon. So a realist statement about the independence of the world from thought must have some qualifications. Here is my formulation:

  Commonsense Realism: We all inhabit a common reality, which has a structure that exists independently of what people think and say about it, except insofar as reality is comprised of, or is causally affected by, thoughts, theories, and other symbols.

  The realist accepts that we may all have different views about the world and different perspectives on it. Despite that, we are all here living in and interacting with the same world. Let us now return to issues involving science.

  12.3 A Statement of Scientific Realism

  How should scientific realism be formulated? One possibility is to see the scientific realist as asserting that the world really is the way it is described by our best-established scientific theories. We might say: there really are electrons, chemical elements, genes, and so on. The world as described by science is the real world. Michael Devitt is an example of a scientific realist who expresses his position in this way (1997).

  My approach will be different. I agree with Bas van Fraassen, and others, who argue that it is a mistake to express the scientific realist position in a way that depends on the accuracy of our current scientific theories. If we express scientific realism by asserting the real existence of the entities recognized by science now, then if our current theories turn out to be false, scientific realism will be false too.

  Should we worry about the possibility that our best-established theories will turn out to be wrong? Devitt thinks that so long as we do not commit ourselves to realism about speculative ideas at the frontiers of science, we need not worry. Others think that this confidence shows disregard for the historical record; we should always recognize the genuine possibility that well-established parts of science will run into trouble in the future.

  How do we decide this massive question about the right level of confidence to have in current science? My suggestion is that we don't decide it here. Instead we should separate this question from the question of scientific realism. A scientific realist position is compatible with a variety of different attitudes about the reliability of our current theories. We want a formulation of scientific realism that is expressed as a claim about the enterprise of science as a whole.

  One complication comes from the following question: must the scientific realist also be a commonsense realist? Is it possible-in principle-that science could tell us that commonsense realism is false? The problem is made vivid by puzzles with quantum mechanics, one of the basic theories in modern physics. According to quantum mechanics, the state of a physical system is partially determined by the act of measurement. Some interpretations of quantum mechanics see this as causing problems for commonsense realist ideas about the relation between human thought and physical reality. These interpretations of quantum mechanics are very controversial. Like a lot of other philosophers, I have been quietly hoping that further work will eventually show them to be completely mistaken. But that is not the point that matters here. The point is this: should we allow for the possibility that science could conflict with commonsense realism? If we say that scientific realism does assume commonsense realism, we seem to be committed to holding on to an everyday, unreflective picture of the world, regardless of what science ends up saying. But if we sever scientific realism from commonsense realism, it becomes hard to formulate a general claim about how the aim of science is to describe the real world.

  My response to the problem is to modify commonsense realism so that it allows for the possibility of unexpected, uncommonsensical relations between thought and reality at large. Commonsense realism as previously formulated allowed for the possibility of causal links between thought and the rest of reality. It is often hard to tell whether a connection posited by science is a causal connection or not. So let us widen the class of relations between thought and the world that realism accepts; science might add new cases. Because we are modifying commonsense realism to make it more responsive to science, this is a naturalistic modification.

  Commonsense Realism Naturalized: We all inhabit a common reality, which has a structure that exists independently of what people think and say about it, except insofar as reality is comprised of thoughts, theories, and other symbols, and except insofar as reality is dependent on thoughts, theories, and other symbols in ways that might be uncovered by science.

  Once we have made this modification, it is reasonable to include commonsense realism as part of scientific realism. Here is my preferred statement of scientific realism:

  Scientific Realism:

  i. Commonsense realism naturalized.

  z. One actual and reasonable aim of science is to give us accurate descriptions (and other representations) of what reality is like. This project includes giving us accurate representations of aspects of reality that are unobservable.

  In this sense, I am a scientific realist.

  Several comments on this formulation are needed. First, clause 2 says that one aim of science is to represent the structure of the world. Nothing implies here that this is the only aim of science. There might be other aims as well. And some particular theories-even whole research programsmight be developed in a way intended to serve other purposes.

  Second, I said "actual and reasonable aim." The first part of this is a claim about the goals behind at least a good proportion of actual scientific work. The second part claims that scientists are not deluded or irrational in making this their goal. They can reasonably hope to succeed at least some of the time.

  But I do not say how often they succeed. No part of my statement of scientific realism endorses our current particular scientific theories. In some areas of science, it's hard to imagine that we could be badly wrong in our current views; it's hard to imagine that we could be wrong in believing that tuberculosis is caused by a bacterium and that chemical bonding occurs via the interactions of outer-shell electrons in atoms. Still, my statement of scientific realism is intended to capture the possibility of both optimistic and pessimistic versions. An optimistic scientific realist thinks we can be confident that science is succeeding in uncovering the basic structure of the world and how it works. The pessimistic option is more cautious, even slightly skeptical. A pessimistic scientific realist might be someone who thinks that it is very hard for our feeble minds to get to the right theories, that evidence is often misleading, a
nd that we tend to get too confident too quickly.

  So there is a range of possible attitudes within scientific realism toward our chances of really understanding how the world works. Although there is a range, there is also a limit. My statement of scientific realism says that giving us accurate representations of the world is a reasonable aim of science. If someone thought it was just about impossible for us to get to the right theories, then it is hard to see how it could be a reasonable aim of science to try to do so. So there is a limit to the pessimism that is compatible with scientific realism as I understand it; extreme pessimism is not compatible. I think of Popper as someone who is getting close to this limit but who does not actually reach it.

  Although Kuhns most famous discussions of realism are his notorious claims about how the world changes when paradigms change, at other times he seems more like a pessimistic scientific realist. These are passages where Kuhn seems to think that the world is just so complicated that our theories will always run into trouble in the end-and this is a fact about the world that is independent of paradigms. We try to "force" nature into "boxes," but nature resists. All paradigms are doomed to fail eventually. This skeptical realist view is more coherent and more interesting than Kuhn's "changing worlds" position.

  Much of the recent philosophical debate under the heading "scientific realism" has really been discussion of whether we should be optimistic or pessimistic about the aspirations of science to represent the world accurately (Psillos 1999). Some hold that fundamental ideas have changed so often within science-especially within physics-that we should always expect our current views to turn out to be wrong. Sometimes this argument is called the "pessimistic meta-induction." The prefix "meta" is misleading here, because the argument is not an induction about inductions; it's more like an induction about explanatory inferences. So let's call it "the pessimistic induction from the history of science." The pessimists give long lists of previously posited theoretical entities like phlogiston and caloric that we now think do not exist (Laudan 1981). Optimists reply with long lists of theoretical entities that once were questionable but which we now think definitely do exist-like atoms, germs, and genes.

  These debates only have the ability to threaten scientific realism of the kind defended here it if they threaten to establish extreme pessimism. They do not support extreme pessimism. But the debates are interesting in their own right. What level of confidence should we have in our current theories, given the dramatic history of change in science? We should not think that this question is one to be settled solely by the historical track record. We might have reason to believe that our methods of hypothesizing and testing theories have improved over the years. But history will certainly give us interesting data on the question.

  We might find good reason to have different levels of confidence, and also different kinds of confidence, in different domains of science. Ernan McMullin (1984) has rightly urged that we not think of the parts of physics that deal with the ultimate structure of reality as a model for all of science. Basic physics is where we deal with the most inaccessible entities, those furthest from the domain our minds are adapted to dealing with. In basic physics we often find ourselves with powerful mathematical formalisms that are hard to interpret. These facts give us grounds for caution. And where we are optimistic, we might have grounds for optimism about some features of our theories and not others. McMullin and also John Worrall (1989) have developed versions of the idea that the confidence we should have about basic physics is confidence that low-level structural features of the world have been captured reliably by our models and equations. That is a special kind of confidence.

  All those factors that are relevant in the case of fundamental physics do not apply in the case of molecular biology. There we deal with entities that are far from the lowest levels, entities that we have a variety of kinds of access to. We do not find ourselves with powerful mathematical formalisms that are hard to interpret. The history of this field also supports a view holding that we are steadily accumulating knowledge of how biological molecules work and how they operate in the processes of life. So trying to work out the right attitude to have toward molecular biology is not the same as trying to work out the right attitude toward theoretical physics.

  Realists sometimes claim that there is a general argument from the success of scientific theories to their truth. It is sometimes claimed that realism is the only philosophy of science that does not make the success of science into a miracle (Smart 1968; Putnam 1978). This line of argument has been unimpressive as a defense of realism. The real world will definitely have some role in affecting the success and failure of theories. Theories will do well or badly partly because of their relations to the world in which they are used and investigated. But there are many kinds of ways in which the link between theory and reality can generate success, especially in the short or medium term. Accurate representation of the world is not the only way. Theories can contain errors that compensate for each other. And theories can be successful despite being very wrong about the kinds of things they posit, provided they have the right structure in crucial places. Here is a simple example used by Laudan: Sadi Carnot thought that heat was a fluid, but he worked out some of the basic ideas of thermodynamics accurately despite this. The flow of a fluid was similar enough to patterns in the transfer of kinetic energy between molecules for his mistake not to matter much. Realists need to give up the idea that success in science points directly or unambiguously toward the truth of theories.

  I hope my reasons for setting things up in the way I have are becoming clear. Much of the literature has held that scientific realists must be optimistic about current theories and about the history of science. I resisted that formulation of the issue. There is no point in arguing too much about the term "scientific realism," but there are benefits from organizing the issues in the way I have. What I call scientific realism is a fairly definite yesor-no choice. (Fairly definite; see section 1z.7.) This is also a choice about fundamental philosophical issues. The question about the right level of optimism to have about well-established scientific theories is not a question that has a simple answer that can be easily summarized. There we need to distinguish between different scientific fields, different kinds of theories, different kinds of success, and different kinds of optimism. In many cases we surely have good reason to be optimistic, but simple slogans should not be trusted.

  One more comment on my formulation of scientific realism is needed. I said that science aims to give us "accurate descriptions and other representations of what reality is like." This is meant to be very broad, because there are lots of different kinds of representation used by different sciences. Some philosophers think that the main goal for a realist is truth; a good theory is a true theory. So they might want to formulate realism by saying that science aims to give us true theories. But the concepts of truth and falsity are only easy to apply in cases where a representation is in the form of language. In addition to linguistic representations, science often uses mathematical models, and other kinds of models, to describe phenomena. A scientific claim might also be expressed using a diagram. So I use the term "accurate representation" in a broad way to include true linguistic descriptions, pictures and diagrams that resemble reality in the way they are supposed to, models that have the right structural similarity to aspects of the world, and so on. I will return to these issues in the final section of this chapter.

  12.4 Challenges from Traditional Empiricism

  Scientific realism is now a popular position, but it has faced constant criticisms and challenges. Many of the most influential philosophers have thought that there is at least something wrong with scientific realism of the kind described in the previous section. Let's do a quick survey of the philosophers discussed so far in this book. Logical positivism was mostly opposed to scientific realism. Kuhn was vague and not always consistent, but he mostly opposed it. Many sociologists of science have certainly opposed it, including Latour. Goodma
n, the inventor of the "new riddle of induction," was opposed to it. Van Fraassen, who influenced my statement of what scientific realism is, rejects the view. So does Laudan. Feyerabend is hard to assess. Popper is in favor of scientific realism. Many of the naturalists discussed in the two previous chapters are scientific realists (including Fodor, Hull, and Kircher), but not all are.

  The critics listed above do not agree on what is wrong with scientific realism. I will divide the various forms of opposition into three broad families. Critics of realism differ among themselves just as much as they differ from the realists.

  First, scientific realism has often been challenged by traditional forms of empiricism. In this book I will defend both scientific realism and a kind of empiricism, but this is not always an easy alliance. Indeed, one side of the debate about realism is often referred to as a debate between realism and empiricism.

  Traditional empiricists tend to worry about both common sense and scientific realism, and they often worry for reasons having to do with knowledge. If there was a real world existing beyond our thoughts and sensations, how could we ever know anything about it? Empiricists believe that our senses provide us with our only source of factual knowledge. Many empiricists have thought that sensory evidence is not good enough for us to regard ourselves as having access to a "real world" of the kind the realist is committed to. And it seems strange (though not absurd, I think) to be in a position where you simultaneously say that a real world exists and also say we can never have any knowledge about it whatsoever.

  The logical positivists recast these issues in terms of their theory of language. In the heyday of logical positivism, traditional philosophical questions about the "reality of the external world" were regarded as meaningless and empty. So the logical positivist attitude to most discussions of the "relation between science and reality" is that no side of the debate is saying anything meaningful and the whole discussion is a waste of time.