1. Introduction

“Had Eddington lost his faith in humanity, the (amount of) gravitational deflection would not have been observed”. This statement is about events that did not happen. The antecedent is against the facts – it is counterfactual. The whole statement is a counterfactual conditional that describes how things would have gone, had something gone differently. For simplicity, we will use the term “counterfactual” to refer to counterfactual conditionals. Historical counterfactuals, then, are counterfactuals that are relevant in historiography.

This post addresses the functions of counterfactuals in historiography and the principles that govern counterfactual reasoning. The questions at hand are how counterfactuals function in historiography and how we can reason about them in a controlled manner. However, there is a dilemma when it comes to historical counterfactuals. On one hand, they serve several important functions such as establishing contingency and grounding explanations. On the other hand, their epistemology and motivation have been constantly questioned. Some have dismissed them as “parlour games” (Carr 1961, 97) while others have criticized them for lacking proper evidential support and failing to distinguish between different levels of causation (Evans 2013, 148). In the following discussion, we will explore the functions of historical counterfactuals and use this as a basis to consider their epistemology and motivations.

2. Historical Counterfactuals

As we already noted, counterfactuals are conditionals where the antecedent is false. They tell us what (supposedly) would have happened in the case that the antecedent happened. The important questions are (i) how, exactly, the antecedent is supposed to have happened, and (ii) how we connect the antecedent to the consequent.

We shall discuss the questions in detail in Section 4. Here we can note that many ways in which the antecedent might have occurred give us the wrong verdict about the counterfactual. For example, in most contexts, the claim “Had the barometer reading dropped, a storm would have occurred” is simply wrong. If the barometer reading is supposed to drop because the atmospheric pressure drops, then the counterfactual is true but we find it misleading. There is something wrong with the type of change in the scenario. Moreover, in some cases, changing the antecedent might be meaningless and the whole counterfactual requires an interpretation. For example, if we say “Had I been a dog, I would have eaten the same food all the time”, there is the question of what would it even mean to change me into a dog.

The second question is a difficult one because we cannot have, by definition, any direct evidence about counterfactual scenarios, i.e., how things would have developed, had something been different. This problem has often led to a dismissive attitude towards historical counterfactuals and the idea that they are mere “parlour games”. The fact that philosophical literature on counterfactuals has not been all that interested in the epistemic details does not help the issue.

Now, to put some flesh on the bones, consider an illustration.

First, consider the following: In 1919, Arthur Eddington confirmed Albert Einstein’s theory of relativity by observing gravitational deflection during a solar eclipse. Born into a Quaker family, Eddington aimed to reunite the international scientific community, fractured during WWI. He appreciated Einstein’s pacifist background and sought to bridge the gap between British and German science. Einstein published his special theory of relativity in 1905 and later developed the general theory of relativity, predicting that gravity should bend light to a specific degree. Eddington became intrigued by this prediction and its potential to reconnect the scientific community. He persuaded British scientists of the theory’s importance and organized an expedition to observe the solar eclipse. After analyzing the data from photographs taken during the eclipse, Eddington declared that the predicted deflection was confirmed. This observation not only validated Einstein’s theory but also demonstrated the importance of international scientific collaboration and made Einstein an scientific star.

We can ask many questions about counterfactual developments of the scenario. For example, we could ask, what would have happened, had Eddington not been a Quaker. Would the observation of the deflection have been made in 1919? In a sense, it seems that the observation would not have been made, had Eddington not been a Quaker. His religious background gave him an unique incentive in British science to focus on Einstein’s theory. However, it seems a bit unclear what we mean by the antecedent. Do we assume that Eddington had some other religious beliefs or that he had lost his faith in humanity? These two scenarios lead to vastly different outcomes. Moreover, how can we track the scenario where Eddington did not have the religious beliefs? We can see that many interesting issues arise with respect to counterfactuals.

The example above illustrates how historical counterfactuals arise from our knowledge about historical events. The example appears to be rather controlled – it does not involve wildly speculative reasoning. Moreover, while there might be some obscurity in the antecedent, the problems of understanding the starting points of the relevant scenarios do not seem overwhelming. We will contrast these observations with philosophical problematization in Section 4. Before that, we will look at what counterfactuals can be used for.

3. The Functions of Historical Counterfactuals

Historiographical counterfactuals provide insights that are relevant to our historiographical understanding and knowledge. I will identify three main functions: (i) providing insights on the actual course of history, (ii) providing insights on contingency, (iii) providing explanatory understanding. I will point out that counterfactuals are able to satisfy functions (i) and (ii) through function (iii).

First, it can be argued that counterfactuals are useful enhancements for historical reasoning about the actual course of events. In general, one could say that counterfactual thinking is a good exercise to open up historiographical imagination and leads to novel insights. However, this does not answer the questions of why there should be any epistemic credibility in the scenarios thus constructed (we can open up our imagination by using all sorts of intellectual games) and how, exactly, opening up imagination serves sound historiography.

Counterfactual thinking has been examined in detail in terms of its relationship with the study of actual history. According to scholars like Nolan (2016), counterfactual thinking enables historians to recognize how history might have unfolded differently and identify the paths that best fit the available evidence. Some scholars claim that counterfactual histories can also help identify actual causal chains by exploring how changes in events would have influenced the outcome. However, counterfactual reasoning is typically considered just one method among others for establishing actual causal chains. Scholars like Tucker (2016) and Woolf (2016) suggest that counterfactual scenarios can be used as a heuristic tool to establish actual causal sequences, but they should ultimately disappear once the historical interpretation is complete.

The problem is that, as we will see below, counterfactuals are much more intimately connected to causal explanations in historiography than the perspective above suggests. It is impossible to have an explicit causal explanation in historiography that cites only an actual sequence of history. As Bulhof puts it, “Counterfactuals, causes, and explanations are three sides of the same strange three-sided coin; you cannot have one without the other two” (1999, 147). This is good news for counterfactuals: Given that even those who remain skeptical (like Evans 2013; 2016) about historical counterfactuals still allow there to be causal explanations in historiography, we can see that we can’t escape from historical counterfactuals.

Secondly, it has been suggested many times that counterfactuals can reveal contingency. In Ben-Menahem’s suggestion, historical necessity and contingency should not be associated with causation and randomness. Rather, both notions are causal notions. They are defined in terms of degrees of sensitivity to initial conditions and intervening factors. (1997; 2016; see also Virmajoki 2018.) The idea is that the more sensitive the outcome is to the changes in the initial conditions and intervening factors, the more contingent it is. What connects counterfactuals and contingency in these types of analysis is the idea that contingency is revealed by estimating how easily things could have been different.

Finally, counterfactuals are central to causal explanations in history. That there is some connection between causal explanation and historical counterfactuals is almost universally agreed (see Weber 1949; Bulhof 1999; Maar 2014, 2016; Ben-Menahem 2016; Sunstein 2016; Bunzl 2004; De Mey & Weber 2003; Lebow 2000). However, as we have seen, some philosophers have questioned whether counterfactuals are essential to explanations in historiography. Next, we see that they are.

In order to understand the issue, we can notice that some have associated historiographical explanations with contingency. It has been argued that (causal) narrative explanations are especially good at exhibiting contingency. Narrative explanations explain particular events via causal sequences concluding with the explanandum (Currie 2014, 1165). Beatty has suggested that “narratives are especially good at representing contingency and accounting for contingent outcomes” (2016, 34; see also Beatty and Carrera 2011). A narrative provides understanding about turning points in a sequence of events leading to the outcome (Gallie 1964; Beatty 2016).

But what makes narrative explanations explanations, i.e., how are things connected to each other in an explanation? Whatever the answer, it must be able to capture the central role of contingencies in narrative explanations.

There are two relatively recent candidates that are highly relevant to the issue at hand. One is the so-called mechanistic account of explanation. According to this account, historiographical explanations describe ephemeral mechanisms that produce the historical outcome that we wish to explain. An ephemeral mechanism is a short-lived and non-stable collection of interacting parts whose interactions can be characterized by direct, invariant, change-relating generalizations (Glennan 2010, 260). What is important for us is that the mechanistic account of explanation requires that the change-relating generalizations used in explanations are “counterfactual supporting” (Glennan 2010, 257). In other words, if we cannot rely on counterfactual considerations, we cannot construct historiographical explanations.

The other candidate is a counterfactual account of explanation. As the name promises, counterfactuals are central to explanations in this account. Counterfactual accounts of explanations have roots deep in the history (e.g., Weber 1949), but one of the most popular accounts is the so-called interventionist account. This account was developed by Woodward (2003) and applied to historiography by Virmajoki (2022). In this account of explanation, “the underlying or unifying idea in the notion of causal explanation is the idea that an explanation must answer what-if-things-had-been-different questions, or exhibit information about a pattern of dependency” (2003, 201). Explanations provide information about counterfactual dependencies between explanans and explanandum and are contrastive in nature.

Given the interventionist account of explanation (and counterfactual accounts in general), counterfactuals are at the core of historiographical explanations. Counterfactual considerations are not a mere tool to establish actual causal sequences that explain. In fact, it does not even make sense to say that we could explain through actual causal sequences. A description of a sequence does not include information about patterns of dependencies. Only if we grasp these patterns, we have achieved understanding. We need counterfactual scenarios in order to have historical explanations; having knowledge of counterfactual scenarios (patterns of dependence) is identical to having an explanation.

This means that those who believe that historiography can provide causal or explanatory insights must accept that historical counterfactuals make sense and can be tracked.

Given the central role of counterfactuals in historiographical explanations, we can make sense of how we can gain insights into contingencies and the actual history through counterfactuals.

First, given that contingency is related to the sensitivity of the outcome on other factors, contingency can be assessed on the basis of explanations. Given that an explanation provides us with information about counterfactual situations where the outcome would have been different, we can assess the contingency by asking how plausible it is that these situations occurred. Narrative explanations are especially good at representing contingency and accounting for contingent outcomes because explanations in general are suitable for this task.

Counterfactuals provide insights into actual history because they require assumptions about how the world works, which tells us what we know and assume about the past. Tracking counterfactual scenarios can reveal the generalizations we rely on when interpreting history and guide us in critically assessing those generalizations. Additionally, our inability to track certain counterfactuals can lead to new questions and guide the search for evidence.

We have seen that counterfactuals are central and useful in historiography. They are essential for historical explanations and historical contingency. Moreover, the construction of counterfactual-based explanations is useful for the study and interpretation of the actual history. While this latter function could perhaps be achieved by other means as well, as Maar (2014) argues, it is difficult to see why we should not be omnivores (to use Currie’s [2018] phrase) when it comes to ways of gaining historiographical insights. Now it is time to turn to the epistemological issues surrounding counterfactuals. The discussion already suggests that “the difference between counterfactual and ‘factual’ history may … be marginal”, as Lebow (2000, 553) puts it.

4. Counterfactual Reasoning

There are two main questions concerning counterfactual reasoning in historiography. First, how should we understand the change required in the antecedent? This question has received a lot of attention in the philosophy of historical counterfactuals. Secondly, how can we track what would have happened after the antecedent? This question has received less attention but is more interesting. Yet, we shall proceed by discussing the first question first.

Consider the counterfactual “Had the barometer reading dropped, a storm would have occurred”. It seems that the truth of this claim depends on how the barometer reading is supposed to have dropped. If it dropped because the atmospheric pressure dropped, then a storm would have occurred. On the other hand, if it dropped because I adjusted the needle with my finger, a storm would not have occurred. How should we understand the change in the antecedent – the transition period as some name it? (Greene 2021.)

In his canonical analysis, stemming from philosophical needs, David Lewis suggested that the antecedent is supposed to be changed by a miracle, the smallest change required for the antecedent to occur (1973). There are technical problems in the suggestion (see Woodward 2003, 133–145) but the philosophy of historiography has focused more on the motivational problem. It has been argued that historical counterfactuals need to be understood in terms of changes that were historically plausible or coherent. In Elster’s terms, the antecedent “must be capable of insertion into the real past” (1978, 184).

This line of thought has been common in the literature (see Greene 2021). The thought is expressed in the so-called minimal rewrite rule which says that we should avoid far-fetched counterfactuals in historiography and consider only those counterfactual situations that could have happened, given the actual history (see Tetlock and Belkin 1996, 23-25). There have been several attempts to make the rule more precise by identifying some categories of events whose occurrence supposedly were likely or plausible.

One of the suggestions has been that we should focus on decisions and scenarios that were considered by the people involved in the historical situations. “We should consider as plausible or probable only those alternatives which we can show on the basis of contemporary evidence that contemporaries actually considered” (Ferguson 1999, 86). In a similar spirit, Tetlock and Belkin suggest that “investigators might agree to constrain counterfactual speculation in a host of more specific ways: by considering as antecedents only those policy options that participants themselves considered” (1996, 23).

However, it seems attempts like these to tell categorically which events are likely or plausible do not lead us anywhere. Notice that the criterion above is obviously false. What paths the historical actors considered does not track what really is plausible. At the end of the 19th century, fundamental changes in physics were not widely considered plausible. We know that the fundamental changes were very much plausible.

An even more serious problem is that these types of considerations confuse the original counterfactual “Had C been the case, E would have been the case” and the counterfactual “Had D been the case, E would have been the case”, where D is the most plausible conditions where C was the case. That we are interested in the dependence between C and E, the dependencies between D and E are irrelevant. And if we assess the dependencies between D and E, we are not drawing conclusions about dependencies between C and E.

The whole question of what antecedents are plausible seems completely misleading.

Notice that there are perfectly legitimate ways to define how the antecedent was changed. We can use the notion of a surgical intervention from Woodward. When we make an intervention to change the actual C* to counterfactual C, we change the system in such a way that only C* is changed to C. The intervention does not have an independent effect on other relevant factors (Woodward 2003, 14; see p. 98 for details). This means that when we consider a counterfactual, we consider a case where the antecedent holds because of this type of stipulated change. We are not asking how the change could “really” have happened. We are interested in what follows from the change from C* to C. This is what our counterfactual “Had C been the case…” is about. Of course, one can have an idea of how the antecedent could “really” have occurred. In this case, one can stipulate this change in the history and asks what would have happened. However, again, this counterfactual is not “Had C been the case…” but “Had D been the case…” where D is supposed to lead to C and something else.

Of course, there might be counterfactuals that do not make sense (see Sunstein 2016, 347). The reason is that, in such cases, we do not conceive any intervention that could inject the antecedent into the history. However, there is no reason why such counterfactuals should provide reasons to doubt counterfactuals globally. “Had Eddington not made the observation, Einstein’s theory would not have become famous when it did” is a perfectly legitimate counterfactual while “Had Bill Clinton been a dry match, would ignite if struck” is not. I see no reason to assume that most important historical counterfactuals fall into the nonsensical category.

The interventionist approach to counterfactuals is useful because counterfactuals are essential to explanation. In order for C to explain E, E must depend causally on C. However, not just any counterfactual change is sufficient for this purpose. For example, if the barometer reading drops due to a drop in atmospheric pressure, this does not explain the storm because the change in atmospheric pressure does not change the reading in the correct way. This is not an intervention, as it has an independent effect on E that does not go through C. Therefore, in order to construct relevant counterfactuals for explanation, we need the interventionist interpretation of the “transition period.” Adopting the interventionist approach is the most natural way to proceed in counterfactual thinking, as it helps us achieve the goal of constructing explanations through scenarios.

There are strengths in this type of “antimetaphysical” strategy of using stipulations that go beyond our needs to explain. Sometimes it seems that the discussion on counterfactual histories is built on the assumption that there are possible events that were “really” the ones that would have taken place if remove something from the history – as if they were in line waiting to take the place of the actual event. Consider the following case. Robert Fogel’s Railroads and American Economic Growth (1964) is widely considered a classic in counterfactual histories. As Bulhof summarizes,

“Using data from the time, and economic theories, Fogel constructed a mathematical model of the economic condition of the United States without the development of the railroad. This uses quantified rules of transformation derived from contemporary economic theories, and from them, Fogel argued for a modal conclusion: had the U.S. not developed the railroad, little would have been different” (1999, 166).

One of the leading critics of counterfactual histories, Richard Evans, writes as follows:

“Robert Fogel, in his analysis of the impact of railway building on the American economy, is not trying to imagine what America would have been like without railways, rather, he is building a statistical model of the US economy with the contribution of railways taken out, in order to show what the actual contribution of railways was. He is not, in other words, saying that the railways might not have been built. He is not constructing an alternative timeline. This is a fundamentally different procedure from the hypotheses of the true counterfactual historian, whose business is precisely to posit what might have been had some event or process not taken place.” (2016, 462.)

It seems that Evans claims that counterfactuals related to causal explanation (“the actual contribution”) are not what we should mean by counterfactual histories. Counterfactual histories, it seems, must assume that the antecedent could have been the case in some very strong sense of “could”. The final sentence is revealing: it seems that counterfactual histories must focus on removing something from the history and seeing what takes its place.

Asking the question “What if the actual C* did not occur?” is a very unfortunate way of asking counterfactual questions. It suggests that we use some sort of a metaphysical vacuum cleaner to suck C* out of the history and see what event from the metaphysical space of possibilities fulfills the vacuum. This is not what is done when someone constructs a historical counterfactual. If it was, then we all should follow the critics of counterfactual histories. Yet, we do not need to follow the critics. We should stick to stipulating what kind of event we presume to have occurred due to an intervention and try to track the scenario through.

What Is Agency?

It is interesting to note that the kinds of problems that have been raised in historiography of human past does not seem to have a counterpart in the historical sciences. When we are asking what would have happened to the Cambrian explosion, had the Snowball Earth been different, we do not have similar tendencies to ponder about the most plausible way to change the Snowball Earth (or how they relate to human decisions, as there are none). Whereas, in the case of human history, one can ask bizarre questions like whether different decisions by an agent would have led to a situation where the historical agents “cease to recognise or to acknowledge themselves as the agents that they were” (Hawthorne 1991, 166; see Greene 2021), one would never ask similar questions in the historical sciences. It seems that our reasoning about historical counterfactuals in the case of the human past is confused by other dimensions of historical understanding. If counterfactuals were as complicated in their nature as the discussion about their use in human history suggests, then we should expect to find similar problems in neighboring fields. We do not find such problems, and therefore it would be best to stick to a straightforward understanding of the counterfactuals, such as the interventionist, and discuss metaphysical issues (such as when would a historical actor still recognizes herself) later and in other contexts.

This leads to the final issue. How can we track counterfactual scenarios?

To be sure, there are different methods that an individual study can use. For example, we already saw how Fogel used statistical methods, data, and economic theories to answer what would have happened, had the railroads not been built. On the other hand, Tucker has argued that “counterfactual historiographic hypotheses are tested just like factual historiographic hypotheses … We suspend belief in a body of evidence that proves that the counterfactual never happened while maintaining all the rest of the evidence. The probability of the counterfactual depends then on the likelihood of the rest of the evidence given a counterfactual hypothesis.” (2004, 229-230).

It seems that counterfactual scenarios require generalizations that are used to connect the antecedent to the later events. These generalizations can have many origins and forms, and it would be important to give them the same attention in the philosophical analysis as has been given to the debate on the nature of transition periods. The problems are (i) what types of generalizations are needed, and (ii) what the relationship between the use of generalizations in the study of actual history and the study of counterfactuals is.

Virmajoki (2023) has suggested that counterfactual scenarios can be tracked once we make assumptions about the causal situation that held in the past and postulate principles that we assume guided the situation. Here we can note that the difficulty of the task varies greatly in different contexts:

In some (easy) cases, it is rather easy to track counterfactual scenarios. For example, the fact that an article inspired (and caused) a further article can be assessed by reading the latter article. We rely on the principle “An article comments on an earlier one only if its authors were influenced by the earlier article”.

In some (intermediate) cases, we need to do much more to establish a counterfactual dependency. For example, to establish the dependency of acceptance of a scientific idea on the age structure of a scientific community, we need to rely on (i) considerations of the credibility of Planck’s principle that “[a new] scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it” (cited from Hull et al. 1978, 718) and (ii) on the prospects of applying that principle to the case at hand which requires demographic data and careful study of the views and discussions within the community.

In some (difficult) cases we need facts, non-trivial principles, and philosophical assumptions about the issues at hand. For example, when we study the dependence of the 20^th-century developments in physics on the works of Einstein (theory) and Eddington (observation), we need to establish (i) facts (e.g. how widespread was the idea of spacetime-curvature in the early 20th century), (ii) principles (e.g. how did international relationships in science work before and after the WW1), and (iii) philosophical assumptions (what is the role of observation in theory-acceptance?).

This means that tracking counterfactuals is not easy and it rarely can lead to an agreement, as the principles and assumptions are quite difficult to assess (Virmajoki 2023). The more general the principles used are, the more room there is for disagreement. However, counterfactual reasoning is possible and can be performed in a controlled manner. The commonly used term “thought experiment” does not do justice to the process. It is not about imagination but about applying generalizations that can be explicated and discussed. The evaluation of the generalizations also provides us with tools to identify wishful or otherwise biased counterfactual histories, the existence of which has been used as an argument against historical counterfactuals from time to time at least since Carr (1961, 97-98).

Finally, we can notice that all historiography requires generalizations and principles. As Lebow (2000) points out, even the actual history is reconstructed from evidence using generalizations and principles – also the reasonability of these works can be assessed on the basis of the inferences. Surely, it is easier to establish a chain of events A-B-C… in the actual history where we can have direct evidence of each event. However, when we claim that the chain is a causal chain, we already suppose counterfactual dependencies and postulate some principles according to which A is the cause of B and so. The only difference between historiography of the actual world and counterfactual reasoning is in the fact that the inference chains are longer in the latter case. When we reconstruct the actual history, we make inferences from sources to past events and inferences about the causal relations between the actual events. In the case of counterfactual events, we make further inferences about what could have happened instead of the actual events. However, all the reasoning requires generalizations and principles. There is no guarantee that counterfactual reasoning is always more difficult than reasoning about actual history.

5. Conclusion

In this post, we have seen what historical counterfactuals are, what their function is, and how we can reason about counterfactual scenarios. I have argued that historical counterfactuals have important functions that force us to resist skepticism towards their use. I also argued that there are confusions about how historiographical counterfactuals should be handled. The philosophical discussion has focused on how the antecedent of a counterfactual can be established. This has led to natural but ultimately unmotivated ideas about the restrictions in acceptable antecedents in historical counterfactuals. I argue that we can simply stipulate the antecedent in terms of interventionism. Finally, I pointed out that counterfactual reasoning can proceed by using causal knowledge in the forms of generalizations and principles. While counterfactual reasoning is difficult and there are several places for disagreement in the process, the reasoning does not differ fundamentally from the more familiar types of historiography.

References

Beatty, John & Carrera, Isabel (2011). When What Had to Happen Was Not Bound to Happen: History, Chance, Narrative, Evolution. Journal of the Philosophy of History 5 (3):471-495.

Beatty, John (2016). What are narratives good for? Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 58:33-40.

Ben-Menahem, Yemima (1997). “Historical Contingency”. Ratio 10 (2). 99–107.

Ben-Menahem, Yemima (2016). “If Counterfactuals Were Excluded from Historical Reasoning”. Journal of the Philosophy of History 10 (3). 370-381.

Bulhof, Johannes (1999). “What if? Modality and history,” History and Theory 38:2, 145–168.

Bunzl, M. (2004). “Counterfactual history: A user’s guide”. The American Historical Review 109. 845–858.

Carr, E. H. (1961). What Is History?. New York: Vintage.

Currie, Adrian (2014). Narratives, mechanisms and progress in historical science. Synthese 191 (6):1-21.

Currie, Adrian (2018). Rock, Bone, and Ruin: An Optimist’s Guide to the Historical Sciences. The MIT Press.

De Mey, Tim & Weber, Erik (2003). “Explanation and Thought Experiments in History,” History and Theory 42:1. 28–38.

Evans, Richards (2016). “Response”. Journal of the Philosophy of History 10 (3). 457-467.

Ferguson, Niall (1999) Virtual History: Alternatives and Counterfactuals. Basic Books.

Elster, John. 1978 Logic and Society: Contradictions and Possible Worlds. John Wiley & Sons.

Evans. Richards (2013). Altered Pasts: Counterfactuals in History. Waltham, Mass: Brandeis

University Press.

Gallie, W. B. (1964). Philosophy and the Historical Understanding. New York: Schocken Books.

Glennan, Stuart (2010). Ephemeral Mechanisms and Historical Explanation. Erkenntnis 72 (2):251-266.

Greene, Catherine (2021). Historical Counterfactuals, Transition Periods, and the Constraints on Imagination. Hopos: The Journal of the International Society for the History of Philosophy of Science 11 (1). 305-323.

Lebow, R. N. (2000). “What’s so different about a counterfactual?” World Politics 52. 550–585.

Lewis, David (1979). Counterfactual Dependence and Time’s Arrow. Noûs 13 (4):455-476.

Lewis, David K. (1973). Counterfactuals. Cambridge, Blackwell.

Maar, Alexander (2014). “Possible uses of counterfactual thought experiments in history,” Principia: An International Journal of Epistemology 18:1. 87–113.

Maar, Alexander (2016). “Applying D. K. Lewis’s Counterfactual Theory of Causation to the Philosophy of Historiography”. Journal of the Philosophy of History 10 (3). 349-369.

Nolan, Daniel (2016). “The Possibilities of History”. Journal of the Philosophy of History 10 (3). 441-456.

Hull, David L. & Tessner, Peter D. & Diamond, Arthur M. (1978). Planck’s Principle. Science 202 (4369). 717-723.

Reiss, Julian (2009). “Counterfactuals, Thought Experiments, and Singular Causal Analysis in History” Philosophy of Science 76 (5):712-723.

Rosenfeld, Gavriel (2016). “The Ways We Wonder “What If?” Towards a Typology of Historical Counterfactuals”. Journal of the Philosophy of History 10 (3). 382-411.

Soler, Léna & Trizio, Emiliano & Pickering, Andrew (2015). Science As It Could Have Been. Discussing the Contingency/Inevitability Problem. University of Pittsburgh Press.

Stanley, Matthew (2003). ‘An Expedition to Heal the Wounds of War’ The 1919 Eclipse and Eddington as Quaker Adventurer. Isis 94 (1). 57–89.

Stanley, Matthew (2019). Einstein’s War: How Relativity Triumphed Amid the Vicious Nationalism of World War I. Dutton

Sunstein, Cass (2016). “Historical Explanations Always Involve Counterfactual History”. Journal of the Philosophy of History 10 (3). 433-440.

Tetlock, Philip E, and Aaron Belkin (1996) “Counterfactual Thought Experiments in World Politics: Logical, Methodological and psychological Perspectives.” In Tetlock, Philip E, and Aaron Belkin (eds.) Counterfactual Thought Experiments in World Politics. Princeton University Press

Tucker, Aviezer (2016). “Historiographic Counterfactuals and the Philosophy of Historiography. An Introduction”. Journal of the Philosophy of History 10 (3). 333-348.

Tucker, Aviezer (2004). Our Knowledge of the Past: A Philosophy of Historiography. Cambridge University Press.

Virmajoki, Veli (2018). Could Science Be Interestingly Different? Journal of the Philosophy of History 12 (2).

Virmajoki, Veli (2022). What Should We Require from an Account of Explanation in Historiography? Journal of the Philosophy of History 16 (1). 22-53.

Virmajoki, Veli (2023). “Frameworks in Historiography. Explanation, Scenarios, and Futures”. Journal of the Philosophy of History.

Weber, Max (1949). On the Methodology of Social Sciences. The Free Press of Glencoe.

Woodward, James (2003). Making Things Happen. A Theory of Causal Explanations. Oxford University Press.

Woolf, Daniel (2016). “Concerning Altered Pasts: Reflections of an Early Modern Historian”. Journal of the Philosophy of History 10 (3). 413-432.