The philosophy of science embraces all philosophical questions that arise from reflection on science. Because science is widely believed to be the best available means of acquiring knowledge, and because scientific theories seem to tell us a great deal about the nature and functioning of the world, the philosophy of science overlaps considerably with epistemology and metaphysics.
This section might be titled the philosophy of physical or natural science and complements the section of this guide on the Philosophy of Social Science. Philosophical problems arising from particular theories in physics and biology are dealt with in the Philosophy of Physics and Philosophy of Biology sections of this guide. There are also philosophical problems that arise in the context of detailed examination of important developments in the History of Science and they are accorded their own section.
The most fundamental question for the philosophy of science is ‘what is science?’ (although some philosophers think this question is misleading because there are no characteristics which all the things we call science have in common). Other important questions include: Is there a single method by which all the sciences progress? What indeed is a scientific theory? What degree of belief should we have in scientific theories? What is the relationship between theories in the different sciences? Further details of these and other issues are given below.
Each of these books has material on all the major issues in philosophy of science:
Each of these anthologies has relevant readings for all the topics below:
People, including scientists, often talk about the scientific method, but what is it and indeed is there a single method to science at all? Is scientific theory change rational?
Much of the discussion in the philosophy of science concerns the relationship between theory and evidence. We usually assume that if a theory predicts some result for a particular experiment and that result is then observed, then the observation is positive evidence for the truth of the theory and the theory is confirmed by it. However, the problem of induction (see epistemology) is that no amount of evidence for a particular universal generalisation, say that all swans are white, is inconsistent with the piece of evidence refuting it, say the next observed swan being black. Popper famously tried to avoid this problem by proposing that theories are never confirmed by evidence only falsified; in so far as a theory has not yet been falsified, despite our best efforts, then we have reason to carry on using it but we should never think that it has been inductively supported. Others have argued that the problem of induction reappears for Popper’s theory of scientific methodology (which is called falsificationism), and that we cannot do without a notion of theory confirmation. Any theory of confirmation must avoid the so-called paradoxes of confirmation which arise if we adopt an obvious theory of confirmation that fits out intuitions. One celebrated account of confirmation which focuses on the quantifying the degree to which a particular piece of evidence supports a particular theory is Bayesianism.
See also the reading above under Scientific Method. Most introductory books on the philosophy of science have a chapter or two on falsificationism and a good explanation of Popper’s views can be found in W. Newton-Smith, The Rationality of Science, chapter 3.
Is science unified? Can biology and chemistry be reduced to physics? What would reduction consist in if it were possible? These issues have important implications in the philosophy of mind, where the question of whether or not the mental is reducible to or supervenes upon the physical is of great importance, and in metaphysics, since we need not include entities that are reducible in our ontology of basic objects.
Scientific theories, when taken literally, often seem to refer to and describe unobservable entities such as electrons, protons and spacetime geodesics. Should we believe that such things really exist or are they perhaps merely useful fictions like the average taxpayer or the equator? Scientific realists think that we ought to believe in the existence of the unobservable entities postulated by our best scientific theories and that those theories are approximately true. Instrumentalists believe that theories are merely a means by which to codify and predict the phenomena, or that which can be observed. Reasons for scepticism include the problem of the Underdetermination of Theory by Evidence (see below), the past record of empirically successful theories that are now believed to be false, and the difficulty in articulating a precise notion of approximate truth. (There are also special problems for scientific realism in the philosophy of physics.)
A notorious problem in the philosophy of science is that an individual hypothesis in a scientific theory on its own will not imply anything of an observational nature, and so cannot be subjected to empirical testing without being combined with a whole host of other hypotheses, background assumptions and auxiliary hypotheses about initial conditions, the operation of the apparatus and so on. This leads Quine to advocate his confirmational holism according to which any theoretical hypothesis may be retained in the face of any refuting evidence whatsoever provided enough changes are made in the rest of the theoretical system. For van Fraassen the underdetermination argument is an important reason for adopting antirealism about science.
Bayesian and other theories of confirmation have to offer a solution to this problem so C. Howson and P. Urbach, Scientific Reasoning: The Bayesian Approach, P. Horwich, Probability and Evidence, and C. Glymour, Theory and Evidence, all have sections on it.
Science is often thought to tell us about the laws of nature. But what is a law of nature and do they differ from generalisations that happen to be universally true, like no gold spheres are bigger than the earth? (no plutonium spheres are bigger than the earth is an analogous generalisation which is supposed to be law-like). An important (Humean) view of laws (defended by Frank Ramsey and David Lewis) is that they are in fact just a particular sort of regularity. Others hold that laws of nature are relations between universals.
Philosophers of science who are sceptical about the traditional emphasis on laws in the analysis of scientific theories include Bas van Fraassen, in his Laws and Symmetry, part I, and Nancy Cartwright, in her How the Laws of Physics Lie.
What does it mean to give a scientific explanation of something? Presumably such an explanation must involve a scientific theory somehow but can we give necessary and sufficient conditions for something to be an explanation? The orthodox view of explanation was the ‘covering-law model’ but that is now widely contested and various alternative accounts of explanation have been proposed. Important questions that such accounts must answer include: do generalisations in explanations need to be laws of nature?; are all proper explanations causal?; are functional explanations teleological and if so would that make them unacceptable in science (see also Philosophy of Social Science, Philosophy of Biology).
According to the so-called ‘received view’ which is the logical empiricist model of science (most developed by Carnap), theories are essentially linguistic entities consisting of a syntax of theoretical and observational terms, and containing correspondence rules linking theoretical terms with the results of observations. This view of theories is no longer tenable for various reasons but there has been no single successor.
Scientific theories are supposed to be based on observation (see ‘Empiricism’ in Epistemology). However, many philosophers of science have pointed out that observations are guided by and described in terms of theories raising the question of whether or not the former can provide a neutral foundation for scientific knowledge (see ‘Foundationalism’ in Epistemology). It was thought by some logical positivists that language could be divided into ‘observational’ and ‘theoretical’ terms. This view is no longer taken seriously and it is accepted by most philosophers of science that all language is to some degree ‘theory-laden’, but can we even distinguish between observational data and our interpretations of it?