Thinking through Historical Cases

In this post, I discuss problems that are associated with case studies in the philosophy of science and possible solutions to the problems. I also discuss the relevance of the problems to our thinking about science and future.

Preliminary note on methodology

Before going any further, we need to notice that the notion of “case study” has somewhat different meaning in the philosophy of science than in other disciplines. James McAllister writes: “Case studies in the proper sense, we may remind ourselves, are circumscribed exploratory studies that aim to yield insights for a broader category” (2018, 246) while in the philosophy of science a “case studies” include any research that attempts to draw some kind of conclusion from a historical episode of science. McAllister (2018) has criticized the sloppiness in philosophers´ terminology and methods. While there is a lot to learn from this criticism, we have to remember that

  • It is not clear what the nature of a claim of philosophy of science is (i.e. whether they are descriptive, normative, critical or whatever) and therefore it is not clear what methodology is best suited to justify those claims.
  • Philosophical questions are fundamental, and the answers depend epistemic/conceptual/metaphysical schemes. Therefore, it is not clear how to apply clear methods on them. As McAllister admits “constructing quantitative estimates of the degrees of referential and observational success of past scientific theories [something that McAllister argues is relevant in debate concerning scientific realism] is more difficult than harvesting biographical data (2018, 254). It indeed is more difficult, and one could argue that the main aim of philosophy of science is to study what the notions like “degree of referential and observational success” mean and how they could be understood.
  • In relation to (I) and (II), it could be argued that it is not beneficial to get stuck with some limited set of (kinds of) methods as some questions might be, by their nature, such that those methods are not fruitful in answering them.

I will argue that (III) is an especially useful attitude when it comes to the scenario-work. I will argue that the philosophical debate on case studies (in philosophers´ sense) will reveal fundamental problems that we face every time we attempt to use a historical insight in our attempts to understand the world. The problems do not concern any specific method but every possible attempt to understand the past in relation to the present and the future. On the positive side, I will show how philosophers´ understanding of case studies might be useful in estimating possible futures.

The problems

In their excellent paper, “Negotiating History: Contingency, Canonicity, and Case Studies”, Agnes Bolinska and Joseph D. Martin discuss problems that are associated with the case studies in philosophy of science and suggest a solution to the problems. This article is the most systematic account of the problems thus far (as far as I am aware). The problems come in two categories, methodological and metaphysical.

The methodological problems

Construction Bias

We do not have a direct access to the past. Our knowledge is mediated by historiographical accounts. The construction of those accounts is based on many choices, trends, and assumptions. The problem is that there can be different historiographical accounts of some past episode and it is not often obvious which one we should accept. Given that different accounts can support different philosophical theories, it is far from obvious how we could choose between the theories on the basis of history.

As a solution, Bolinska and Martin suggest that we can evaluate different constructions from metahistorical perspective. We can investigate whether the explanatory factors suggested by each account are plausible given what else we know about the history of science and the case at hand. While I think that such “global” approach is essentially correct (see Virmajoki 2019, sec. 6.6), the details need some serious attention. Moreover, even if could, in principle, choose between historical accounts on the basis of global considerations, applying these considerations could be a difficult task. We need to keep in mind the intricacies of the construction bias in our mind when using history as evidence.

Selection bias

The problem is that we can select those historical cases or historiographical studies that support the view we hold. It is easy to cherry-pick cases that support the view while ignoring others.

This problem is not too serious, as cherry-picking only reflects poor research and can be balanced by the more extensive use of cases. However, there still the question of relevance: How do we decide which set of cases is relevant? I have discussed this problem in this post.

Interpretation bias

The problem is that we do not build but we can also interpret case studies by using philosophical theories. For example, we might interpret Richard Bellon´s (2011) work on Darwin either as (a) highlighting the importance of social values in science, or (b) highlighting the value of novel empirical insights (see Virmajoki 2019, Ch. 7). These are both plausible suggestions and the choice between them can only be based on deep philosophical considerations. If we can interpret historical case studies in terms of our preferred philosophical theories, then it is difficult to use those case studies to tell which theory to choose.

Even though Bolinska and Martin point out that “Just because it is possible to select and interpret case studies in light of one’s philosophical aims doesn’t mean one must do so” (p. 40) it is unclear how this help in the theory-evaluation. Even if we could produce the interpretations in a neutral way, the existence of different interpretations still makes it possible to use the same case study to support different philosophical theories. I do not find it plausible that different interpretations only arise when some interpretations are based on explicit bias. For example, in Bellon’s study on Darwin, the two interpretations are almost explicit in the original study (see Virmajoki 2019, Sec. 7.3).

Application bias

“Even if we agree on the facts about historical cases and how to interpret them, we might still disagree about what we ought to conclude on that basis” (p. 39). For example, we might know that scientists rejected a theory T and accepted T* once the experiment E was performed. Then we might ask whether this confirms falsificationism. Someone argues that it does because falsificationism is in play every time a theory is rejected when contradictory evidence is found. On the other hand, someone else argues that falsificationism is not confirmed as the theory T was rejected only when there was a better alternative T*. This person could argue that surely scientists accept the theory that is more compatible with the evidence, but this dimension of theory-comparison can be understood in non-falsificationist terms.

These problems have the following structure: We have a philosophical theory P and some historical case which shows H. The support that H gives to P depends on whether we accept further theoretical framework or interpretation F. Once the structure is explicit, we can easily see that the support P gains from H does not depend on P itself but on F. There is no reason to think that we could not evaluate F independently of P. Surely, the more theoretical commitments there are in our historical interpretations, the more difficult it is to come to a clear conclusion on the basis of historiographical studies. However, this should not be a surprise. Philosophical views are, by their nature, networks of theoretical commitments.

As a summary, we can notice that the use of historical cases in “testing” or thinking about science requires understanding of the theoretical commitments that mediate our access to the past. As a remedy, we have to make those theoretical commitments explicit (something I have discussed here) and we have to study the history “globally” in order to find (or construct – I come back to this later) the relevant patterns.

The metaphysical problems


“Scientific concepts, experimental methods and standards, and even the notion of science itself shift from one historical moment to another” (2020, 50). If this is true (and this indeed seems plausible, see Virmajoki 2019, Ch. 2) then there are no cases of science against which we can test our theories. If there is no science as we know it in the past, then there is nothing to compare our theories with. One could suggest that this is nitpicking: Academic historians may be pedantic about the use of categories, but for the purposes of philosophy of science, it is sufficient that there are past episodes that were scientific enough.

There are problems with this (Virmajoki 2019, 55-56):

First, the number of cases available for the philosophy of science becomes much larger once we allow that philosophical theories can be evaluated against “scientific enough” historical episodes. It seems that one could find support for any imaginable philosophical theory from such vast resources.

Secondly, the notion of relevant similarity seems to guarantee that there cannot be a genuine counterexample for one’s theory. If such a counterexample is suggested, one can always claim that the example is not a relevant historical episode. Because there is not a fixed set of historical scientific episodes, there is nothing that a theory of science must be able to capture. In a sense, a philosophical theory can never be incorrect. And if a theory cannot be incorrect, it cannot be correct either: if there were not scientific episodes in the past (in the strict sense), then there is nothing that a philosophical theory could capture.

However, I do not think that this makes it impossible to have theories of science. Of course, if we think that science is different from other forms of human life, then the scarcity of historical cases makes it difficult to formulate theories of science. However, if we accept that science shares many features with other aspects of human life and knowledge, then we can use whatever knowledge we have about social and cultural worlds to understand science. We could learn a lot about epistemological practices by studying the past and use this knowledge to make science understandable. The idea is that we do not attempt to find theories of science but theories that can be applied also on science.


“To any philosopher seeking to found normative claims about science in history, the critic can respond that any historical episode, or set of episodes, was either chancy or depended upon factors tangential to the philosopher’s normative aims. The critique might be formulated as follows: Because history is not governed by strict, deterministic rules (at least not rules that limited beings such as ourselves can discern), it might in some meaningful sense have gone differently. Because history might have gone otherwise, we have ample reason to doubt whether historical examples can constitute firm evidence for philosophical claims that seek to generalize about scientific practice and process.” (Bolinska and Martin 2020, 41).

The problem of contingency misses an important detail: Even if history could have gone differently, it does not follow that we cannot detect causal or normative patterns. Only if different histories would have led to a similar or equally preferable outcome as the actual history, the actual pattern is irrelevant. Eating healthy foods makes me feel better even if it was possible that I only ate potato chips. Surely, I might not have found out that broccoli makes one feel good if I only ate chips, but this is irrelevant to the existence of the causal pattern. (See detailed discussion about the topic in this post).

The metaphysical problems suggest that we have to seriously rethink the notions of “science”, “history” and “explain” in the idea that philosophy of science can be evaluated in terms of how well it explains the history of science.


Bolinska and Martin suggest that the metaphysical problems can be avoided by focusing on canonical cases.

“[T]o establish a case as canonical with respect to a philosophical aim, we necessarily invoke causal-dependence contingency, which consists of three related, but distinguishable claims:

(1) The causal claim—a causal connection exists between a historical outcome and a particular antecedent factor;

(2) The counterfactual claim—that antecedent factor might plausibly have been different;

(3) The sensitivity claim—the historical outcome was non-robust with respect to changes in that antecedent factor”. (2020, 43.)

In other word: “A canonical case is one that can be explained by factors relevant to the philosophical question at hand—and in which the outcome is sensitive to those explanatory factors. Establishing a case’s canonicity, then, requires demonstrating that those explanatory factors are not washed out by other contingencies inherent in the process.” (2020, 43.)

I completely agree. A philosophical account P that describe a past episode E is acceptable only if the outcome of E depended on factors suggested by P (See Virmajoki 2019, Sec. 3.6).

For example, Darwin´s evolutionary framework was accepted because it produced novel empirical insights. Had it not produced those insights­­­­ it would not have been accepted. This case is canonical with respect to explanatory the power of novel empirical insights. In contrast, the case is not canonical with respect to the status of moral values of science. While Darwin was, at first, condemned in moral terms, a change in his moral status would not have changed the acceptance of the evolutionary framework (see Virmajoki 2019, Sec. 7.3 discussing the issue).

While Bolinska and Martin succeed in analyzing the structure of a case study that show the explanatory relevance of certain types of factors and in explaining what kind of contingency supports a philosophical aim, there are still problems. First, the problem of Heraclitianism is dismissed too quickly as a version of a problem of contingency:

“We can understand Pitt’s Heraclitianism as implying a claim of this sort: history is unsuited for supporting philosophical claims because it could have been otherwise, even given indistinguishable starting parameters.” (2020, 42.)

This would indeed be a suspicious view on the nature of historical processes (2020, 42) but Heraclitianism can be understood simply as an empirical observation: Given what we have learned from the history of science, it seems that sciences in different eras and in different contexts are different.[1] Given this interpretation, the problem is that even if find a canonical case, the results of the case are not generalizable. Surely, a case that can be explained by certain factors is a more salient example of a philosophical theory than the cases that cannot be thus explained (2020, 46) but all this seems to do is to motivate the philosophical theory that does the explaining.

The second problem is related to the first one. In order to establish canonicity, we have to establish claims about causal dependencies. However, if we cannot compare many similar cases (due to Heraclitianism) it is difficult to justify claims about the causal dependencies. I say difficult, not impossible: Making sense of history requires a global survey of principles, regularities and singular contexts that can be fitted together. The problem with the notion of canonicity is that it attempts to explicate “what features a case study would need to have in order to be suitable for drawing inferences about philosophical claims” (2020, 41) but this requires, as we saw, understanding of causal dependencies (a good case study tells how the outcome depends on the antecedent factors). A case study can never enable us to draw inferences without global considerations. One would like to have a notion of canonicity that involves a thin notion of philosophical relevance; a notion that defines a set of cases that philosophical accounts should be able to clarify and defines the set in a way that does not require that a particular philosophical theory in fact explains them. By studying the set globally, we could draw inferences that are based on an exemplary set of scientific cases.

Finally, it is interesting to note how methodological and metaphysical problems affect each other. Let us remind ourselves that the methodological problems concerned the interpretation and use of historical evidence and the conclusion that are drawn from historical studies. There seem to be no Archimedean point for evaluating an interpretation of history. This is not a devastating conclusion as such points are not found in any discipline. To overcome the methodological problems, we have to have a balanced study of different theories and historical cases. However, the metaphysical problems seem to question the fundamental premise of the philosophy of science: that there is something general to be said about principles and dynamics of science. It might be that the methodological problems cannot be solved because different historical cases of science require different interpretations. How do can we know this? By studying the history. Is it easy to make conclusions about the history? No because we have the methodological issues!

I suggest that we stop doing metaphysical-epistemology here and turn to other direction. Can we understand the value of philosophy of science in different terms than suggested by the language of “philosophical theories and explanations” and can we turn the methodological problems into a strength?

Rethinking philosophy of science

First, even if are not able to tell, on the basis of historical studies at least, how science in general works[2], the philosophical efforts have not been futile. At least we know that there is no theory that is able to explain all of it. This is relevant to our thinking about the future. We should not follow any single theory in construction of scenarios but create taxonomies of the future on the basis of many theories. I have discussed this idea in a previous post. Moreover, we are able to tell which theories does not work, at least in many cases. For example, falsificationism does not have much historical support even though it is used in popular discourse about science.

Secondly, while we do not know the patterns of good scientific development on the basis of history, we can still attempt to construct such patterns. This requires that we abandon the idea that philosophy of science is mainly a descriptive practice. We can have a continuum of positions with regard to the strength of the construction. At one extreme, we could have a theory that is based on a priori ideas on how knowledge must be sought. At the middle, we have theories that rationally reconstruct history in order to tell how science could have been rational (this idea is discussed here). The idea is that, even if the historical episode was contingent on a factor that is irrelevant from the perspective of a philosophical theory, the historical episode supports the philosophical theory as long as it is true that “had the irrelevant factor been removed and had only the factors cited by the philosophical theory been at work, the outcome would have been the same”. At the other “extreme” there are theories that make “recommendations about what one ought to mean by various [scientific] claims, rather than just attempting to describe how we use those claims. It recognizes that [scientific] claims sometimes are confused, unclear, and ambiguous and suggests how these limitations might be addressed.” (Woodard 2003, 7.) Woodward is referring to causal and explanatory claims, but the point generalizes. A philosophical study of what scientists do can reveal unclarities in those practices. It is no wonder that a neat philosophical theory does not apply to them, historically speaking. However, it can often be useful to formulate a neat account precisely to remove the unclarities that have been stratified historically.

Thirdly, we could follow the spirit of Causal layered analysis (see my discussion here) and reveal deep worldview commitments behind surface phenomena (Inayatullah 1998). Historical philosophy of science can be seen as a practice that attempts to understand the connection between ideas of science (e.g. the realists’ commitment to the approximate truth of the mature theories) and the historical reality. The possible mismatch between theories and history is not a theoretical failure but an insight on how our philosophical views track the historical patterns. Moreover, we can study the past and see how philosophical ideas have been presented in scientific practices and how much effect they have had on those practices. This kind of research can reveal a lot about the relationship and mismatch between human practices and the conceptions associated with those practices.

All three points above appreciate the methodological and metaphysical problems introduced above. First, different historical accounts or interpretations can be seen as a future-oriented strength. They enable us to see the uncertainties and different ways of organizing practices. Moreover, the range of interpretations reveals that our conception of ourselves and our past is deeply affected by theoretical assumptions whose origins are not completely clear (if they were, there would be no methodological problems in the philosophy of science). Secondly, the contingent and changing nature of science can be turned into an insight: We can fight wrong kinds of contingencies once we recognize them and we can prepare different ways to organize research if we understand the transient nature of current practices.

General lessons for the possible futures

Historical cases have an important role in our thinking, at least as heuristics. Often, we might not even recognize that our thinking is based on a historical case. For example, when we discuss about the possible result of a forthcoming election, we often cite previous election and what we know about them. Sometimes it is thought that a long series of elections in our own countries provides a good starting point for our thinking. Sometimes we look at the countries that share some seemingly relevant similarity with ours etc. This kind of thinking-in-terms-of-historical-cases reminds what is going on in the philosophy of science.

Of course, social sciences have their own methodology of thinking through historical cases and there is nothing I can add to that in this post. There is nothing new even in the suggestion that historical cases and data can be understood and interpreted in many ways. This is a constant complication in social sciences. What I want to do is to suggest how the discussion about the methodology of philosophy of science can make more explicit the structure of thinking-in-terms-of-cases and suggest new ways of creating future scenarios.

First, it is important to understand that every historiographical account has what was called “biases”, i.e. the accounts are based on more or less implicit principles that guide the choice of materials, the plotting of the story and the interpretative choices. In some easy cases these might not cause much problems. For example, we can easily check when a traffic law changed. However, in some cases the effect of those assumptions can be trickier. For example, the decisions of the Supreme Court are used in legislative crafting. In this way, those decisions affect our lives. However, the Court has its own principles for deciding the cases. Even though we can assume that those principles are sound ones, we can still note how a concrete event (or set of events) is interpreted through a set of principles and transcribed into a decision that affects our lives through legislation. The principles of the Supreme Court are relatively transparent but it easy to conjecture that other actors that make decisions do not learn from concrete cases through similarly explicit processes. This has two consequences: 1: We have to be as transparent as we can in our principles of interpretation in order to produce epistemically tractable future scenarios. We can use the categories of methodological problems (see above) as a heuristic guide when assessing the transparency and theoretical and other commitments of our principles. 2. We have to understand how different actors (organizations, individuals etc.) think about historical cases in their future-oriented actions. For example, even if we cannot estimate the future legislation, we can find clues from the decisions of the Supreme Court because we know that the legislators use these decisions in their work.

Secondly, it might be fruitful every now and then to view our theoretical systems as tools of thinking rather than attempts to describe the reality as it is. In analog with the points of the previous section, we can:

  1. Accept many different theories (or simply “views” if “theory” is too thick a concept) on a subject matter and provide future taxonomies on the basis of those theories. This leaves open the possibility that one theory is better confirmed than others and its future-structure more likely than others but still we can understand the different possible structures of the future (see this post).
  2. Understand our theories as suggesting different ways of organizing our practices and activities. They might not be descriptively accurate but normatively insightful. Moreover, we should find theories that are reasonably linked with the reality while still telling how things would proceed in idealized (desirable or undesirable) situations.
  3. Understand the thinking-in-terms-of-historical-cases as a measurement of the distance between our conceptions of how things work and the reality. Thinking in terms of historical cases can reveal intellectual and historical stratification in our conceptions and theories. Moreover, comparing our understanding of historical cases to the way the previous generations understood those cases can reveal unrecognized developments in our understanding about the world. This development can be based on the growth of knowledge or simply of fads. In either case, we can learn a lot of ourselves.


Bellon, Richard (2011). Inspiration in the Harness of Daily Labor: Darwin, Botany, and the Triumph of Evolution, 1859–1868. Isis: A Journal of the History of Science 102:393-420.

Bolinska, Agnes & Martin, Joseph D. (2020). Negotiating History: Contingency, Canonicity, and Case Studies. Studies in History and Philosophy of Science Part A 80:37–46.

Inayatullah, Sohail (1998). “Causal Layered Analysis. Poststructuralism as Method”. Futures, Vol. 30, No. 8, pp. 815–829

McAllister, James W. (2018). Using History as Evidence in Philosophy of Science: A Methodological Critique. Journal of the Philosophy of History 12 (2):239-258.

Helloween (1988). Keeper of the Seven Keys Part II.

[1] Remember how Kepler attempted to provide a priori foundations for Copernican system in terms of five regular solids.

[2] I do not mean that philosophers of science think that all sciences are similar; a philosophical theory is “general” as long as it aims to explain a type of research in context-independent way.