Monday, September 14, 2015


216. Theory, concept and fact

In debates in the philosophy of science, it has been claimed that the notion of empirical testing is problematic because observations and facts are ‘theory laden’. One always looks at the world through the glasses of some prior theory, or concept, or mental category. I have gone along with that. Now I take a closer look.

If facts are theory laden, how can theory be tested empirically? Isn’t there circularity involved? There is, of course, if facts are based on the same theory as the one that is tested. But one can distinguish between concepts underlying facts, and theories tested on them. This distinction between concepts and theory was made by the French philosopher Canguilhem[i].

Here, concepts are based on different, earlier theory, now taken for granted and embodied in instruments and methods of observation.

For example, a microscope embodies theory of light, and yields data on which one can test biological theory. Later atomic theory yielded the basis for electronic microscopes that can look ‘deeper’.

Later in this blog I will criticize mainstream economic theory, but I will maintain useful concepts from economics, such as, for example, economies of scale, substitution between factors of production, entry barriers to markets, and transaction costs.

There are levels of factuality. Suppose one has a cloud of data, in some two-dimensional space. One theory might draw a straight line through the cloud, another a curved line, and they fit equally well. They give different interpretations of the cloud but they see their own curve as the ‘fact’ represented by the cloud. Then they might agree about the cloud but not on what the cloud means. One theory may forbid curvature of the line through the cloud while the other demands it. 

An economist might construe the cloud as ‘evidence’ of individual rational choice, a sociologist as ‘evidence’ of imitative herd behaviour. They may accuse each other of ‘not making sense’. They may also try to go back to the cloud and look behind the data to investigate them more closely, or to collect additional data aimed at settling the difference of interpretation, if they can agree on the method of observation.

‘Lower level’ theory underlying concepts used for observation may be more widely shared, even among people adhering to rival ‘higher level’ theories. But the lower level is not infallible or eternal. There regularly are discoveries that alter concepts underlying ‘the facts’.

Hopefully, in a conflict between rival theories one will sooner or later find facts explained by one theory but ‘forbidden’ by the other. It may happen that one theory can explain a fact that the other cannot. There are two views of light: as a stream of particles and as a wave. The wave explains phenomena of interference that the particles do not, but the particles explain phenomena that indicate that light has mass, which the wave does not. So, a dual theory remains.

The problem with social sciences is that such ‘crucial experiments’ seldom exist. There, observation more often remains a matter of perspective, which often hardens into dogma or even ideology.


[i] Gary Gutting, 1989, Michel Foucault’s archaeology of scientific reason, Cambridge U. Press.