In the philosophy of science, theories by Popper, Lakatos, Kuhn, and others have attempted to explain the process of scientific development and the criteria for theory selection, but each has its limitations. Popper’s disprovationalism emphasizes the importance of disproving theories but has practical limitations. Lakatos tried to complement this by proposing the concept of a research program for scientific progress, but Charmus points out that a deeper distinction needs to be made between the impact of researchers’ choices on scientific progress.
Philosophy of science seeks to answer questions such as what is science and what is not science, what methodologies scientific researchers can follow in the development of science, if any, and by what mechanisms the history of scientific progress can be explained. In modern philosophy of science, Popper’s disproversialism, Lakatos’s research program, and Kuhn’s paradigm theory are representative attempts to answer these questions. Each of these approaches offers an interpretation of science from its own perspective, and while they may successfully answer some questions, they have enough criticisms that they are subject to debate and revision. Lakatos’ research program is considered to be a continuation of that approach, addressing some of the problems of disprovism. However, it also has methodological limitations. I noticed that Lakatos’s description of his research program and the limitations pointed out by Charmus in Contemporary Philosophy of Science did not include anything related to the researcher’s confidence in the theory, which is why many attempts at disproving theories have failed. In this article, I will briefly describe the preceding theories, suggest possible modifications to Lakatos’s research program, and then justify them.
The central idea of Popper’s disproversialism is that empirical evidence cannot prove a theory to be true, but it can prove it to be false: even if a theory explains a number of observations, it is still possible that future observations will 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 the failure of attempts to disprove it, which is always tentative. Evidence that a theory is conclusively wrong, rather than reaffirming what is already known, is more informative and drives scientific progress. According to disprovationalism, scientific progress is made when a hypothesis thought to be true is disproved, or when a hypothesis thought to be false fails in an attempt to be disproved.
However, disprovationalism is limited by the fact that observations that can disprove a theory also have their own presuppositions: the accuracy of the experiment, the reliability of the observational instrument, and the truth of the theory underlying the observation. Because these premises can be wrong, just like theories, disprovals of theories are only partially provisional, and it is not uncommon in history for observations that cannot be explained by a particular theory to be considered anomalies or observational errors rather than leading to a disproval of the theory. These are cases that cannot be explained by the “discard theory upon disproof” perspective advocated by disprovism.
To address these issues, Lakatos and Kuhn examined the historical development of science and came to the conclusion that scientific progress should be understood as a “structure” rather than as individual hypotheses. That is, certain parts of a theory are considered fundamentally true by researchers, so that observations that are inconsistent with the theory are not seen as a source of problems, but other parts are subject to revision as research progresses and observations that are inconsistent with predictions emerge. This revision must be independently verifiable. For example, in the case of the Neptune discovery, when Uranus did not follow the orbit predicted by Newtonian mechanics, scientists explored the possibility that the assumption that no other bodies existed outside of Uranus was wrong, rather than that Newtonian mechanics itself was wrong.
Lakatos’ research program views a theory as a collection of hypotheses, with the ones that are considered fundamentally true called the “solid core” and the others called the “guardrails. Researchers follow two guidelines for modifying the guardrails so that the solid core cannot be disproved. One is the negative discovery method, which requires the nucleus to remain unprovable, and the other is the positive discovery method, which modifies the nucleus to explain observations and predict new phenomena. Kuhn’s paradigm is similar to a research program, and advances in Kuhn’s period of normal science correspond almost equally to advances within Lakatos’s program.
The difference between Lakatos and Kuhn is in the mainstreaming of research programs or paradigm shifts. Kuhn describes paradigm shifts in historical context, including interactions between groups of scientists. Lakatos, on the other hand, evaluates mainstream research programs based on the likelihood of success in making new predictions. Research programs that grow into the mainstream because they make new predictions will replace those that fail to make predictions. This can be viewed as a loose form of disproof, where a single disproof does not mean the theory is discarded, but continuous failure to disprove it does.
The question is how long to maintain a research program that fails to make new predictions. For example, the precession parallax predicted by Copernicus was not measured until the 19th century. It is entirely possible that the guard of regressive research programs could be modified to make them more progressive. After all, it is only in hindsight that we can judge one research program as superior to another.
In Contemporary Philosophy of Science, Charmus points out the limitations of viewing Lakatos’s research program as a methodology that scientific researchers consciously follow. It is necessary to distinguish between the rules of theory change that explain the development of science and the norms that scientists actually follow. Lakatos’s research program has been proposed as a norm of discovery, but Charmus sees limitations in this explanation.
He bases his criticism on the following points First, Lakatos fails to provide criteria for selecting between competing research programs. The judgment that a research program is incremental can only be made in retrospect, but it is not a criterion for researchers to choose the program they are currently working on. Second, science cannot be said to have progressed as a result of scientists following Lakatos’s methodology before his theory was proposed.
Charmus suggests modifying Lakatos’ research program by separating researcher selection from theory change. By making sure that the choices of scientists do not directly cause theory change, he is able to explain theory change independent of the choices of researchers. He introduces the notion of the degree of output to express the potential for a research program to evolve. The degree of productivity of a particular research program is unknown to the researchers, so it cannot provide clear norms for theory selection, but it can explain scientific progress by assuming that programs with higher degrees of productivity will gradually become dominant within a group of scientists over time.
I do not believe that Lakatos’s methodology requires us to abandon the researcher selection explanation. The limitations of Lakatos’s theory are refutable. It is true that Lakatos does not provide norms for theory selection, and it is true that the power of a theory is not known at the time of research, but these limitations can be overcome because theory selection does not have to be the right choice for a successful research program.
As Charmus argues, this is a recently devised methodology, but if scientists can describe a common, conscious flow of thought that they follow, it can be a methodology for explaining researcher choice. A guide that includes the scientist’s thought process for selecting a successful theory would make trust in the hypothesis that has withstood more attempts to disprove it an important selection criterion.
It is reasonable to trust scientific theories that have withstood multiple attempts to disprove them over a longer period of study than those that have not.