How does Lakatos’ research program explain the progress of science?

Lakatos’s research program complements disprovism, but it has limitations, and while Charmus has modified it to account for changes in scientific theory, trust in hypotheses that have withstood more disproof is still important.

 

Philosophy of science seeks to answer questions such as what is science and what is not science, whether it is possible to determine the methodology that scientific researchers follow as science develops, and if so, what it is, and by what mechanisms the history of science’s development can be explained. In modern philosophy of science, Popper’s disproversialism, Lakatos’s research program, and Kuhn’s paradigm are representative attempts to answer these questions. Each of these perspectives offers a view of science that successfully answers the previous questions, but also has enough criticisms that it is subject to debate and revision. Lakatos’s research program is said to bridge the gap to some extent, while at the same time addressing some of the most frequently pointed out problems of disprovism. However, there are methodological limitations. From Lakatos’s description of his research program and Charmus’s discussion of its limitations in Contemporary Philosophy of Science, I noticed that it does not include anything related to the researcher’s confidence in the theory, which has been the subject of many failed attempts to disprove. In this article, I will briefly explain the above, and in light of this, I will propose and justify another modification of Lakatos’s research program.
The central idea of Popper’s disprovationalism is that empirical evidence cannot prove a theory to be true, but it can certainly prove it to be false. Even if a theory explains a set of observations well enough, there is no guarantee that future observations will not contradict the theory, but if the observations are different from what the theory predicts, we know that the theory is wrong. The basis for scientists’ acceptance of a theory is that attempts to disprove it have failed, and that acceptance is always provisional. Proof that something is conclusively untrue tells us more about the world than reconfirmation of what is already known, and this is what leads to scientific progress. Therefore, in disprovationalism, scientific progress occurs when a hypothesis that was predicted to be true is disproved, or when attempts to disprove a hypothesis that was predicted to be false fail.
However, observations that can disprove a theory also require that the hypotheses that led to the observations (such as the accuracy of the experiment, the accuracy of the observation instrument, and the truth of the theory on which the observations are based) are true. These hypotheses can turn out to be false, just like the theory itself, so both the approval of the theory and the disproof of the theory are provisional. In addition, if an observation is made that cannot be explained by a theory, it is not uncommon in history for the observation to be considered an anomaly or an error in observation rather than leading to a disproof of the theory. These are historical events that cannot be explained from the perspective of disprovationalism, where disprovals lead to the discarding of theories.
Lakatos and Kuhn chose to examine the historical development of science and came to the common conclusion that some theories should be thought of as structures, meaning that some parts of the theory are more fundamental and are taken as fundamentally true by the scientists working on the theory and are not thought to be the source of the problem, even if observations that are inconsistent with the theory exist, while other parts are subject to revision as research progresses and observations that are inconsistent with predictions emerge. However, these revisions must be clear enough that they can be independently verified. In the case of Neptune’s discovery, when Uranus did not follow the orbit predicted by Newtonian mechanics, researchers did not assume that the basic principles of Newtonian mechanics were wrong, but rather that the assumption that there were no other bodies outside of Uranus affecting its motion was incorrect, considering other factors that could lead to discrepancies between theory and observation.
Lakatos’ research program can be thought of as a collection of hypotheses that make up a theory. In the preceding structure, the hypotheses that are considered fundamentally true are called the solid core, and the rest of the hypotheses are called the guardrails. A scientist working on a research program considers a solid core to be unprovable by decision, not by argument, and modifies the guardrails to protect the core from disproof if observations appear that contradict the theory. The researcher works with two guidelines: negative discovery, which requires that the rigid core not be subject to disprove, and positive discovery, which requires that the rigid core be modified to explain certain observations and predict new phenomena. Kuhn presents his paradigm in terms of a research program, and the advances within the preceding research program correspond almost equally to the advances in Kuhn’s period of normal science.
Where Lakatos and Kuhn differ is in the change of research programs/paradigms that become mainstream. Kuhn describes paradigm shifts in more detail and with historical context, including the interactions between groups of scientists and not just rationality. Lakatos, on the other hand, thinks in terms of how many new predictions can be made: research programs that enable new predictions to emerge, at least occasionally, grow into the mainstream, while those that consistently fail to make predictions are discarded. This can be thought of as a loose form of disproof: a single disproof does not immediately cause a theory to be discarded, but if it fails to account for persistent disproofs, the theory is discarded.
The question is how long does it take to fail to make new predictions before a research program is abandoned? The precession parallax predicted by Copernicus in the 16th century was not measured until the 19th century. It is always possible for a regressive research program to develop a new modification to its armor that makes it progressive again. In the end, it is only with the benefit of hindsight that one research program can be judged to be better than the other competing programs.
In Contemporary Philosophy of Science, Charmus suggests that there are limitations to thinking of Lakatos’s research program as a methodology consciously followed by scientific researchers. It is one thing to be able to explain the historical development of science in terms of rules, but it is another to be able to determine the norms that researchers actually thought about and followed at the time of the development of science. The former explains the process of theory change, while the latter explains the choices that researchers made that led to that change. Lakatos’ research program and the negative/positive discovery method have been proposed as methodologies to explain researchers’ choices in the sense of the norms within which discoveries are made, and Charmus points out the limitations of such explanations.
Chamus’s point is based on the following. First, as with the problem described above, Lakatos does not provide a criterion for choosing between competing research programs. It cannot explain researchers’ choices between research programs, since it is only possible to determine which research programs are more progressive by looking back in time. Furthermore, while it is true that scientists before Lakatos’s theory was proposed led to the changes in science that Lakatos’s methodology describes, it is not true that they consciously followed Lakatos’s methodology, which was only recently devised.
Charmus suggests modifying Lakatos’ research program to account for changes in scientific theory that are not the result of researchers’ choices. In Lakatos’s program, theory change is the direct result of researchers’ choices, and he does not distinguish between theory change and choice, but he fails to account for researchers’ choices among theories, and the resulting explanation of theory change is incomplete.
Charmus acknowledges that he cannot provide norms for researchers’ choices, but he separates the two so that researchers’ choices do not directly lead to theory change, allowing him to explain theory change independently of providing norms for choice.
He introduces the notion of degree of power, which expresses the amount of different possibilities a theory can develop. The degree of power of a theory is not immediately known by the researchers working on it, nor is it necessary for them to know in order for science to progress. All that is required is the assumption that there are “scientists of a particular talent, skill, and mind for the advancement of science in society,” that the objective opportunities contained in the degree of productivity in a given research program will manifest themselves over a long period of time in the population of scientists being studied. If this assumption is true, then programs with a higher degree of output will gradually prevail over time, regardless of the judgment of individual scientists. Since researchers are not informed about the power of current theories, no norms for theory selection can be given to them. Since any theory can have high power regardless of its progressivity to date, it is also reasonable for scientists to study regressive theories at the time.
But I don’t think Lakatos’ methodology requires us to abandon explanations for scientists’ choices. Each of the arguments that Charmus gives for the limitations of Lakatos’s theory as a methodology can be refuted. It is true that Lakatos does not provide norms for choosing between theories, and as Charmus argues, it is impossible to know the extent of a theory’s power at the time it is studied. However, in Charmus’s discussion above, the choices scientists make and the changes in theories are decoupled, so that the choices scientists make and the rules they follow need not be the right choices for successful theories.
Furthermore, although it is a methodology that has only recently been devised, it can be a methodology for explaining scientists’ choices if it describes a stream of thought that scientists in the past have consciously followed in common. It cannot be a guide to guarantee the success of any individual scientist, but it should contain the scientist’s thinking for choosing a successful theory. I propose confidence in hypotheses that have withstood more attempts to disprove them as one such likely guiding factor.
Among the hypotheses that make up a theory, it is reasonable to trust those that have withstood more attempts to disprove them-that is, those that have been studied longer or are the basis for more successful predictions-than those that have not. To return to the example of the discovery of Neptune, the decision of scientists at the time to follow Newtonian mechanics need not be explained solely as a methodological decision that is difficult to explain. The fundamental hypotheses of Newtonian mechanics include the inverse square law of gravity and the laws of force and acceleration, and many of the predictions derived from Newtonian mechanics are based on these hypotheses in common. The basic hypotheses of Newtonian mechanics have received a kind of historical validation in that they have withstood the attempts to disprove them by the entire field of Newtonian mechanics up to that point. It is therefore reasonable for a practical researcher to consider the possibility that the underlying hypothesis contained in the hard core is wrong, rather than the possibility that the other, protective hypotheses on which the predictions and observations are based are wrong. And so it takes a long period of unresolved, decisive error to dismiss a basic hypothesis.
This idea is unlikely to serve as a basis for choosing between all competing research programs, but it is likely to play a role in judgments between two research programs that are similar in the hypotheses that make up each, but differ in the choice of robust nuclei. That is, there are hypotheses in the structure of some theories that are the basis for more predictions than others, and scientists will tend to choose the research program that has those hypotheses as its robust nuclei based on prior thinking. There is always a chance that an old theory will be disproved, so this tendency can be wrong, but as we have seen, it is not a problem in explaining the development of science.
In this article, we have described Lakatos’s theory, which is a loose adaptation of Popper’s theory to explain historical events, and we have described Charmus’s criticisms and revisions of it. I accepted Charmus’s revised research program, but pointed out that a revised research program that accounts for changes in theory can also suggest common ideas about how researchers choose theories. He agreed with Charmus that it is impossible for a scientist to choose the most successful theory at any given time, but he criticized the idea that scientists cannot consciously follow a methodology when advances in science occur. The impossibility of judging earlier, more successful theories leads to the impossibility of presenting the methodologies that lead individual scientists to success, but it does not lead to the impossibility of presenting the methodologies that individual scientists followed during the period of scientific development. And as one of the guidelines that could be included in such a methodology, he suggested trust in hypotheses that have withstood more disproofs. This discussion is productive because it seeks to explain more of what philosophy of science aims to do, and we can expect to discover or elaborate on these guidelines further, improving our understanding of scientists’ choices.