Introduction: success, change, and empiricism

Scientists pose questions about the world and, in some sense, they succeed in arriving, however tentatively, at answers. In this chapter we ask how we should understand that success. In particular, should we think of scientific inquiry as a project aimed at revealing a true description of the world, even when that description involves things that we cannot possibly see, hear, touch, taste, or smell? Or should we consider scientists to be engaged in a search for theories that are simply in some way useful?

On the one hand, the ways in which scientists are sometimes successful would be, some philosophers argue, inexplicable or miraculous unless it were the case that they have accepted some theories that are (at least approximately) true. The approximate truth of some of their theories is the best explanation of certain successes that scientists have enjoyed. Scientific realism maintains that we should regard empirically successful theories in the mature sciences (like physics, chemistry, and biology) as at least approximately true because this is the best explanation of the success of those theories.

On the other hand, we have seen how theories can change in science in dramatic and unexpected ways. A theory might even seem to enjoy tremendous success, only to be replaced by a theory with a radically different conception of the basic entities responsible for the phenomena in its domain.

Growth through falsification

The philosopher of science Karl Popper considered the problem of induction so grave as to require the elimination of induction from the methods employed in science. He went so far as to assert that “in my view there is no such thing as induction” (Popper, 1992, 40).

In light of our discussion of inductive arguments in the previous chapter, such a proposal may sound outrageous. Surely, it might seem, we need inductive reasoning in science; it is by such reasoning that we confirm or support general propositions such as natural laws. That is how we learn from experience, fostering the growth of scientific knowledge.

Scientific knowledge does indeed grow, according to Popper. Understanding this growth is for him the central problem of the philosophy of science and of epistemology (the theory of knowledge) more generally. But he regarded the characterization of this growth in terms of induction as deeply mistaken. The belief that inductive arguments from the results of particular observations serve to confirm, support, or justify general empirical propositions – a belief we can label ‘inductivism’ – is a dangerous illusion. That inductivism is the wrong way to understand this growth can be seen most clearly, according to Popper, if we consider an alternative view, which he calls falsificationism.

The key to falsification lies in a logical asymmetry that we noted previously.

Introduction

We have seen how Duhem's arguments indicate limitations on how far deductive logic alone can guide us to decisions about scientific theories based on experimental outcomes. In the early decades of the twentieth century, another group of philosophers and scientists on the European continent were pursuing a philosophical program that attempted to exploit the resources of formal logic as far as possible in the articulation of a ‘scientific philosophy.’

In this chapter we will take a brief look at the philosophical movement known as logical empiricism or logical positivism, paying particular attention to a central project of the movement: an account of the structure of scientific theories. The preoccupation of philosophers of science with the question of theory structure has its roots in an underlying commitment – rarely examined or even articulated throughout most of the twentieth century – to the importance of theories as the primary contribution of scientific inquiry to human knowledge. Later, some philosophers challenged this assumption, arguing that experimentation produces knowledge that is philosophically significant beyond simply its role in testing broad explanatory theories (Ackermann, 1985; Franklin, 1986; Hacking, 1983; Mayo, 1996). Nonetheless, the question of theory structure remains highly relevant to our understanding of scientific knowledge, and the fate of logical empiricist views has had important consequences for the subsequent development of philosophy of science, as we will explore in the next several chapters.

Introduction

In this chapter and the next we will consider the contributions of Imre Lakatos and Paul Feyerabend, respectively. They were contemporaries of Thomas Kuhn who shared some of Kuhn's commitments but developed them into strikingly different visions of the scientific enterprise.

Like Kuhn, both Lakatos and Feyerabend emphasize that an adequate philosophy of science has to answer to the history of science (though they disagreed on the exact role history should play). Like Kuhn, they insist that the theoretical framework in which a scientist works influences the way that she interprets her evidence, making the logical empiricists' sharp distinction between theoretical and observational vocabularies untenable.

Yet Kuhn, Lakatos, and Feyerabend offer very distinct philosophies with different aims. An important difference concerns the rationality of science. For Kuhn, an adequate account of revolutionary science must allow for rational disagreement among scientists, but beyond that rationality is not central to his analysis. For both Lakatos and Feyerabend, the problem of rationality is central, but they take positions that are, at least on the surface, diametrically opposed. Lakatos seeks to demonstrate the (at least partial) rationality of the historical progression of scientific thought (and to distinguish it from irrational pseudoscience). Feyerabend, on the other hand, argues that any attempt to constrain science by a philosophical model of rationality (whether that of the logical empiricists, or Popper, or anyone else) will prove counter productive, inhibiting the free development of the individual scientist and blocking the growth of scientific knowledge.

Introduction

We might think of the problem of induction in terms of a logical gap between the premises of an inductive argument and its conclusion. So that scientific knowledge could be achieved without having to leap such gaps, Karl Popper proposed a theory of scientific method that focused on falsification rather than induction. In doing so, however, he had to confront the fact that, although falsifications can be represented with a gapless, deductively valid argument, logic does not require one to make the inference thus represented. Another gap, however, looms over both inductivism and falsificationism.

We saw in the previous chapter how Zig could have avoided falsifying the Law of Doughnut Delectability by insisting that the non-delicious object he tasted was not, after all, really a doughnut. In such a case the logical gap exists because logic does not force on the observer a particular description of one's observation. Another logical gap arises because the logical connection between the hypothesis or theory being tested and the description of the data or observations is not direct. One needs additional premises (often called auxiliary hypotheses or auxiliary assumptions) in order to establish a logical connection between a theory and a description of a possible observation. Falsification starts with a determination of the falsehood of the observation that is expected assuming that both the hypothesis under test and the auxiliary assumptions are true.

Introduction

The last few chapters have presented a series of challenges to the idea that disputes over the status of scientific theories can be resolved unambiguously by appealing to the evidence. We have encountered arguments that purport to tell us that evidence in some way or another underdetermines the choice of theories, that the very meanings of the words used to describe the evidence may incommensurably depend on the theory one chooses when interpreting the evidence, that different scientists will disagree about the value of various features of the evidence, that even when the evidence seems to point strongly against a theory, one may rationally hang on to that theory, and even that for any given body of evidence, opposing conclusions based on incompatible methodological rules can equally advance scientific knowledge.

Taken together, such arguments appear to throw a rather large quantity of cold water on whatever embers of hope may remain for the logical empiricists' project of providing a neutral framework that scientists could use to evaluate the degree to which any given body of data supports or confirms any given hypothesis.

Many contemporary philosophers of science, however, have sought an alternate route by which they might pursue something like this project. Moreover, they have sought to do this in a way that might accommodate some of the post-positivist insights discussed in earlier chapters.