Theory: A word that gets used a lot in discussing science, or attacking it.

Theories are only verified hypotheses, verified by more or less numerous facts. Those verified by the most facts are the best, but even then they are never final, never to be absolutely believed. [Claude Bernard, 1865, Introduction to the Study of Experimental Medicine.]

The term comes from the Greek for "viewing" or "contemplation". It used to be used to mean something along the lines of "laws of nature" + "methods" for a particular domain, or subject.

The term was used extensively from the beginnings of the scientific revolution, but it has a number of different interpretations. Sometimes it meant just a mathematical model of something, as in Galileo's heliocentric model of the solar system (he said "world", but things have expanded since then) or Newton's theory of gravity and forces. Other times a theory was intended to mean something more explanatory, including mechanisms, or actual physical objects that cause the phenomena. This sense is found, for example, in Harvey's theory of the circulation of the blood, where Newton's theory was not mechanical, because the nature of gravity was unresolved in it (he simply named it and described it mathematically, leaving it for later science to explain what it is, which had to wait for modern physics).

Darwin, on the other hand, used the term rather loosely, referring to "my theory", when in fact he had a number of hypotheses. The relationship between "hypothesis", which is usually a specific idea that can be tested in isolation from other claims made in a discipline, and a theory, which is intended to cover the whole domain in question, is not clear in scientific usage, and they often play the same role.

"Theory" has played a significant role in the understanding of science since the early nineteenth century, but was not considered central before then. Instead, philosophers and scientists spoke more often of [Natural] Law, and of hypothesis, defined by Mill as “any supposition which we make (either without actual evidence, or on evidence avowedly insufficient) in order to endeavour to deduce from it conclusions in accordance with facts which are known to be real” (System of Logic III.XIV.4) As philosophical interpretation became more language based and conceptually analytical, however, theory became more important in the analytical tradition of science, and in the Marxist tradition, theory (theoria) was often contrasted to practice (praxis), following Marx in the Theses on Feuerbach.

In the period following Darwin and Einstein, however, “theory” became a term that applied to a set of laws and hypotheses that were well-attested, coherent, precise and covered all of the explanatory domain in question. Hence, Einstein’s general theory of relativity is understood to cover all dynamics previously covered by Newtonian physics, with the added benefits of improved accuracy and precision, and increased coverage of phenomena that Newtonian physics could not explain (see below: “theory reduction”). Theories replaced earlier theories on rational grounds, it was held, in particular by the philosophical school known as the Logical Positivists, or the Vienna Circle, and this replacement was considered absolute progress.

Popper’s account and theories as explanations

Theories were thought at this time to be constructed on the basis of evidence, through logical induction – generalising from observed instances to formulate universal claims. This set up the Problem of Induction which David Hume had first formulated in the latter part of the eighteenth century, and Karl Popper revived as a problem in 1934, in his Logik der Forschung – “the logic of research”, unfortunately titled The Logic of Scientific Discovery when published in English (Popper 1959). Popper observed, as Hume had, that a universal claim cannot be derived from any finite number of observations, and so cannot be rationally justified or verified, as the positivists had sought to do. He defined a theories as “universal statements … nets cast to catch what we call ‘the world’: to rationalize, to explain, and to master it.” (p59) Theories explained events by giving a deduction from their universal laws and singular initial conditions that described the event to be explained.

Carl Hempel generalized this as what came to be known as the Deductive-Nomological Model of explanation (D-N, cf. Hempel 1957). Variations for law-like generalisations that were not exceptionless universal statements, and for stochastic and probabilistic inferences were also developed by Hempel and others. In particular, a historical form of explanation worked, so to speak, in reverse: if you could produce a law-like generalisation from which, together with some initial and boundary terms, you could deduce (or induce or infer in some manner) the events that were recorded, then you had offered an explanation of the events – the so-called Covering Law model (Dray 1957). If there were more than one generalisation that covered the event, then the best theory was the one that covered the most events or cases. Prediction was not possible in historical science, of course, but retrodiction or the “prediction” of already known events from the explanatory covering law explanation was considered to be equivalent logically to prediction in the experimental and present sciences.

Social accounts

Falsificationism, as Popper’s view was called, underwent several revisions under challenges that to falsify a particular hypothesis meant to accept that several ancillary hypotheses had to be accepted, and that a negative result might in fact falsify them instead, known as the Duhem-Quine hypothesis, (Lakatos 1970; Quine 1953). One reaction, by Feyerabend (1975), asserted extreme methodological anarchism under the slogan “anything goes”, basically allowing a scientist to use or assert anything they liked from creationism to astrology in explanation. Feyerabend may have intended this as a gadfly or as a piece of revolutionary thinking. In this, he focused on the political and social nature of science and its theories - they were to be regarded as serving the political and social interests of the wider community.

Later accounts of the nature of theories arose from a reaction to Popper’s falsificationism. Thomas Kuhn, in the tradition of historians of science like Robert K. Merton, proposed that scientific theories acted as global paradigms (Kuhn 1962), a term he later came to replace with several different terms that more exactly expressed the various roles “paradigm” played in his account of science, following criticism (Lakatos 1970). Kuhn proposed science alternated between “normal science” and “revolutionary science”. During normal periods, a theory acted as a guiding principle in which its terms set out the ontology of the domain under explanation. Methods, viable explanations, disciplinary boundaries and even reality itself was specified by the theoretical paradigm (the latter being an extreme over-interpretation of Kuhn’s ideas In short, a theory was an entire worldview, and during revolutions, there was an incommensurability of terms, ideas and methods that in effect made progress and communication difficult if not impossible.

After Kuhn, Imre Lakatos (1970; 1978) proposed that science consisted of “research programmes” which were open to revision at the edges but consisted of a core of protected hypotheses. Unlike Quine’s earlier reviseability thesis, which asserted that nothing was protected from rational revision, Lakatos held that any research programme had a number of unreviseable theses or theories. These acted as the centre of a research programme, which could either be progressive or static. A progressive research programme was a sign of an active science.

One social approach to theory is sometimes referred to as “social constructionism”, the sociology of knowledge, or “externalism”, and derives largely from the so-called Edinburgh School (cf. Bloor 1976; Manicas and Rosenberg 1985). According to this approach, scientific theories are constructed for social and political reasons to support the status quo of the scientific community (Latour and Woolgar 1986). Subsequent ideas about theories in the analytic tradition of the philosophy of science have not been as widely discussed in the social sciences as the ideas of Kuhn and Popper, which have tended instead to follow in the tradition of Foucault, Derrida and other continental philosophers, whose accounts treat theories more as social constructs and ideologies.

Structural and semantic accounts

However, there have still been several projects to make the notion of a theory more exact within the analytic tradition. Attempts were made by several logicians to formalise Kuhnian theory as set theoretical structures (Stegmüller 1977), hence the name Structuralism (not to be confused with the French philosophical movement). Theories were often thought to be at least in part axiomatic systems (Suppes 1967).

A more general and novel attempt to formalise scientific theories was undertaken by Frederick Suppe and Bas van Fraassen (Suppe 1977; Suppe 1988; van Fraassen 1980), whose view of science was based on it being a collection of models that were semantic structures. The Semantic Conception as it became known, treated theories as explanatory conceptual structures in a mode similar to the D-N models, but added that theories were interpreted semantically, so that the model itself was insufficient as an explanation until all the terms had been assigned referents. The semantic view also treated theories as restrictions on the field requiring explanation - to propose a theory was to specify the explanatory domain it was to cover. A similar account of theoretical explanation as being a position held in a contrast space of alternatives was developed independently in the social sciences by Garfinkel (1981). Currently, a theory is regarded in the philosophy of science as a conceptual entity that plays an explanatory role in a science. Rosenberg (1985: 121) defines it as a “relatively small body of general laws that work together to explain a large number of empirical generalizations, often by describing an underlying mechanism common to all of them”. A movement known as the New Experimentalism (cf. Hacking 1983) treats a theory as a summary of empirical observations based upon experimental interaction with physical objects.

The Semantic Account differs from the so-called Syntactic Account in its view of explanation. The Syntactic Account relies somewhat on Covering Law, or Nomological-Deductive, views of explanation. The semantic view, however is an erotetic view, in which explanations are answers to specific questions in a "contrast space" of logical options. Hence, an explanation specifies why the outcome occurs rather than other possible outcomes. As such, law plays a lesser role in explanation, although explanations can also rely on them if they are available.

A recent development is the use of causal explanation (Woodward 2003), in which explanations (some of them, anyway) are supposed to give a causal account of how an outcome was produced. In some senses, this is a revival of mechanistic accounts.

Theory reduction

A theory that “reduces” to another theory is regarded as having been supplanted by that theory (Nagel 1961), and that progress has been made. The reducing theory is expected to have a wider coverage of phenomena or explanatory domain than the theory it reduces. A number of critics have rejected the ideal of explanatory unity of the sciences, particularly in biology (Rosenberg 1994), and deny that reduction is either inevitable or desirable. In particular, genetic reductionism in psychology is regarded as flawed. This parallels the question of methodological individualism in the social sciences (Lachapelle 2000). In the majority of cases where reduction is considered bad, it is because social behaviours have been regarded as reducing to some biological process of disposition, and this is thought to be either politically unacceptable, or flawed due to the importation into the theory of culturally relative presumptions on the part of the theorist (see Segerstråle 2000). Either way, this is a problem that is considered to reach into the nature of science itself.

Theory in Social Sciences

Social scientists usually mean by “theory” a particular theory - a functionalist, structuralist or socioecological theory of social systems (Layton 1997), for example. However, while “social facts” have been defined by Durkheim (1982), along with method, and similar ideas used in other sociological theory traditions, the notion of a theory has been treated as something obvious or self-evident. How to test or formulated theories in general has been assumed or passed by, in general, and the ways in which theories explained their subject have been left unspecified.

The social sciences tend to rely on conceptions of theory building and testing from the physical and historical sciences. Cultural anthropology has often relied upon conceptions taken from critical theory or social constructivism, in which a theory is something constructed either by the researcher or by the community of which the researcher is a part. Mixed conceptions of theory are rare.

Theory and pseudoscience

Those who promote pseudoscientific ideas such as HIV denial, global warming skepticism, creationism and intelligent design, often claim that the ruling view is a bad theory (or "unsound science") because it fails to meet some requirement of a good scientific theory. Usually, this is based on a misunderstanding of Popper's view, or of Kuhn, but sometimes it relies on the social constructionist account of science (in the form of "all theories are constructed for convenience or political reasons, and so you should allow that we, too, have a scientific theory"). Basically, there are several responses, either based on the Demarcation Principle ("science is X, and this is not X"; see here or here) or on the view that science is what scientists do. The latter sounds very question begging, but it need not be if the notion of science is anchored in clear cases, and used to assess borderline cases. For my money, science is something that generates useful, productive theories that advance our knowledge.


A technical article by William Lycan is available


Bloor, David. 1976. Knowledge and social imagery, Routledge direct editions. London; Boston: Routledge and K. Paul.

Dray, William. 1957. Laws and explanation in history. Oxford: Oxford University Press.

Durkheim, Émile, and Steven Lukes. 1982. The rules of sociological method. First American ed. New York: Free Press.

Feyerabend, Paul K. 1975. Against Method. New York: Verso Editions.

Garfinkel, Alan. 1981. Forms of explanation: rethinking the questions in social theory. New Haven, Conn: Yale University Press.

Hacking, Ian. 1983. Representing and intervening: introductory topics in the philosophy of natural science. Cambridge UK: Cambridge University Press.

Hempel, Carl G. 1957. Aspects of scientific explanation, and other essays in the philosophy of science. New York: The Free Press.

Kuhn, Thomas S. 1962. The structure of scientific revolutions, International encyclopedia of unified science; v. 2 no. 2. Chicago: University of Chicago Press.

Lachapelle, Jean. 2000. Cultural Evolution, Reductionism in the Social Sciences, and Explanatory Pluralism. Philosophy of the Social Sciences 30 (3):331-361.

Lakatos, Imre. 1970. Falsification and the methodology of scientific research programmes. In Criticism and the growth of knowledge, edited by I. Lakatos and A. Musgrave. London: Cambridge University Press.

———. 1978. The methodology of scientific research programmes, His Philosophical papers; v. 1. Cambridge; New York: Cambridge University Press.

Latour, Bruno, and Steve Woolgar. 1986. Laboratory life: the construction of scientific facts. Princeton, NJ: Princeton University Press.

Layton, Robert. 1997. An introduction to theory in anthropology. Cambridge: Cambridge University Press.

Manicas, Peter T. , and Alan Rosenberg. 1985. Naturalism, Epistemological Individualism and "the Strong Programme" in Sociology of Knowledge. Journal for the Theory of Social Behaviour 15:76-101.

Nagel, Ernst. 1961. The Structure of Science: Problems in the Logic of Scientific Explanation. London: Routledge and Kegan Paul.

Popper, Karl R. 1959. The logic of scientific discovery. Translated by K. Popper, J. Freed and L. Freed. London: Hutchinson.

Quine, Willard Van Ormand. 1953. From a logical point of view : 9 logico-philosophical essays. Cambridge MA: Harvard University Press.

Rosenberg, Alexander. 1985. The structure of biological science. Cambridge, UK; New York: Cambridge University Press.

———. 1994. Instrumental biology, or, The disunity of science. Chicago: University of Chicago Press.

Segerstråle, Ullica. 2000. Defenders of the truth: the sociobiology debate. Oxford: Oxford University Press.

Stegmüller, Wolfgang. 1977. Collected papers on epistemology, philosophy of science and history of philosophy. 2 vols. Vol. 2. Dordrecht: Reidel.

Suppe, Frederick. 1977. The search for philosophic understanding of scientific theories. In The structure of scientific theories, edited by F. Suppe. Urbana, IL: University of Illinois Press.

———. 1988. The semantic conception of theories and scientific realism. Urbana, IL: University of Illinois Press.

Suppes, Patrick. 1967. What is a scientific theory? In Philosophy of Science Today, edited by S. Morgenbesser. New York: Basic Books.

van Fraassen, Bas. 1980. The scientific image. Oxford: Clarendon Press.

Woodward, James. 2003. Making things happen: a theory of causal explanation. New York: Oxford University Press.


More like this

I have an uncanny ability to offend those who I shouldn't be offending, with bad jokes. In a recent post I put in a Tom Lehrer video where he mocks sociology. Having had philosophy mocked by my friends and contacts over the years (you study what? Your navel?), I guess I am a bit inured to such…
In this post, I want to propose my own view, or rather the views I have come to accept, about the nature of science. [Part 1; Part 2] There are three major phases in the philosophical view of science. The first was around in the nineteenth century - science is the use of inductive logic based…
Rob Skipper has an excellent post at his blog entitled What Scientific Explanation Isn't. It's a good explanation of the DN (deductive nomological) model of explanation offered by Carl Hempel, which has come across some serious criticism of late. By coincidence, this was a major issue in the two…
Lately there has been a rediscovery on the blogia of C. P. Snow's Two Cultures - which initially was the divide between those who understood the Second Law of Thermodynamics and those who don't, but is now, it appears, between those who know math and those who don't, and the respective attitudes.…

John Wilkins wrote

The semantic view, however is an erotetic view...

I wonder how many, like me, misread that to imply that theories are what science finds to be the sexiest explanations...

By Ian H Spedding FCD (not verified) on 17 Feb 2007 #permalink

A better summarization that many full-fledged books I have read. Very well done. Thank you.

By carey allen (not verified) on 17 Feb 2007 #permalink

Extremely well done! Best definition of its length I've ever seen, speaking of someone who's taught many hundreds of students from Kuhn, and thousands more without Kuhn.

Glad you avoided the "Occam's razor" confusion, which is also something abused by pseudoscience.

I suggest that you publish this in hardcopy somewhere refereed, so people who don't believe in blogs can be handed the citation.

If Compression makes one of the 2 bestselling magazines in North America (Reader's Digest), then Expansion must be worthy of publication, in such a fine example.


"One reaction, by Feyerabend (1975), asserted extreme methodological anarchism under the slogan anything goes, basically allowing a scientist to use or assert anything they liked from creationism to astrology in explanation. Feyerabend may have intended this as a gadfly or as a piece of revolutionary thinking."

Poor reading of Feyerabend here. Feyerabend never intended "anything goes" to be take literally, rather it was his reductio ad absurdum. One should not be "overly rigid" in any method they apply or it will lead to absurdity.

"'Anything goes' is not the one and only 'principle' of a new methodology, recommended by me. It is the only way in which those firmly committed to universal standards and wishing to understand history in their terms can describe my account of traditions and research practices ... If this account is correct then all a rationalist can say about science (and about any other interesting activity) is: anything goes." P.K. Feyerabend

By Simon Moon (not verified) on 19 Feb 2007 #permalink