1. Introduction
1.1 The Methodology of Scientific Research Programmes

Two 20th century philosophers of science have strongly shaped how scientists, to this day, view what science is and how it works. The first was Karl Popper, who advocated falsifiability as the defining characteristic demarcating scientific theories from non-scientific ones [1]. The other was Thomas Kuhn, who held that in actual practice scientists are not attempting to falsify their most important theories—or paradigms—but are rather trying to preserve them [2]. Less well known is that the seemingly opposing views of Popper and Kuhn have been synthesized by the Hungarian philosopher Imre Lakatos [3]. This synthesis is called the methodology of scientific research programmes (MSRP).

According to MSRP, scientists are working within so-called ‘research programmes’, which are somewhat similar to Kuhnian paradigms. A research programme consists of a static ‘hard core’ of fixed beliefs, and a dynamic ‘protective belt’ of auxiliary hypotheses and background knowledge (Figure 1). Whenever the programme is confronted with anomalous data, the hard core is shielded from refutation by changing something in the protective belt. There are innumerable ways in which the protective belt can be altered: this may involve the addition of a new parameter, different assumed starting conditions, the recognition of a new type of experimental error, or the proposal of a brand-new hypothesis—whatever is needed to account for the evidence without affecting the hard core.

In addition to a hard core and protective belt, a research programme also has a heuristic. This is a set of (mathematical or experimental) tools and principles that guides researchers as they develop the programme. The heuristic has a negative aspect (which simply says: preserve the hard core), but also a positive aspect. The positive heuristic tells researchers which questions to ask, where to look for interesting data, and how to sophisticate the protective belt in such a way that the programme as a whole can explain increasingly detailed data.

The Popperian element of MSRP surfaces in how Lakatos appraises research programmes. Programmes are assessed by how their protective belts evolve over time. When a research programme undergoes changes, whether it is in response to anomalies or by the forward momentum of the positive heuristic, Lakatos demands that these changes lead to the prediction of novel facts. If they do, the changes are ‘theoretically progressive’, and if the predictions are corroborated, they are ‘empirically progressive’, but if the changes merely accommodate already known facts in an ad hoc fashion, Lakatos calls them ‘degenerative’.

So rather than appraising theories in temporal isolation, under MSRP one must consider the progression of the research programme over an extensive period. If a research programme is characterized by progressive changes, scientists have good reason to continue scientific effort in that direction. If, on the other hand, a research programme generally features degenerative adjustments, scientists have a rational basis for switching to another, more promising research programme.