An approach to searching for scientific progress in philosophy

Abstract

There is much debate about whether philosophy makes progress and if yes, how much. Dellsén, Lawler, and Norton have recently argued that any sensible discussion about this must presuppose some definition of philosophical progress. I do not dispute that this is one possible approach, but in this paper, I offer an alternative and argue that it is viable and promising. We can distinguish two questions about progress in philosophy: one can ask whether philosophy achieves results of the kind that science aims to achieve, scientific progress, or one can ask whether philosophy achieves results of a kind that science does not aim to achieve, philosophical but not scientific progress. Much of the debate, I propose, focuses on the former question, which we can largely address without producing an account of philosophical, or indeed scientific progress. We can empirically evaluate whether philosophy displays the markers associated with scientific progress in all the existing accounts of scientific progress irrespective of which account is correct. This can settle empirical questions about the comparison between philosophy and science even if we do not have a consensus about the analysis of philosophical and scientific progress and on whether philosophy aims at results of a kind that science does not.

The view that philosophy, or at least a part of it, does not make scientific progress has a long history. Kant wrote that metaphysics until his time had been “a mere groping” and took it upon himself to set it on “the secure course of a science” (1781, Bxiv-xv). Despite Kant’s efforts, Russell wrote over a century later that philosophers “have to confess that [their] study has not achieved positive results such as have been achieved by other sciences” (1912, 240).

This view is alive and well today. Among others, Dietrich (2011) argues that philosophy makes no progress, scientific or otherwise, and Beebee (2018) argues that philosophers do not accumulate knowledge, which is the principal aim of science in her view. However, this view also has some detractors. Chalmers (2015) and Wilson (2017) argue that philosophy makes less scientific progress than it would be desirable, but it makes some. Stoljar (2017) argues that philosophy makes as much scientific progress as could be expected, although not quite as much as science. Cappelen (2017) argues that philosophy makes as much scientific progress as science does.

Dellsén et al., (2022) have recently offered a framework to reason about this debate. In their view “any sensible discussion of whether, or the extent to which, philosophy makes progress must presuppose some account of the nature of philosophical progress” (Dellsén et al., 2022, p.832). Their work has greatly improved our understanding of the debate. Dellsén, Lawler, and Norton are certainly right to claim that if questions about the nature of scientific or philosophical progress are ignored, we may not notice when a superficial disagreement about scientific or philosophical progress in philosophy results from an underlying disagreement about definitions. This being said, the observation that questions about definitions must not be ignored does not entail that any sensible view must at least tentatively choose a definition of philosophical progress. In this paper, I present an alternative approach, and I argue that this approach is both viable and promising.

First, I argue that to understand the aforementioned discussion we need not develop novel accounts of philosophical progress. Participants in this debate are asking whether and to what extent philosophy makes the sort of progress that science makes. Accordingly, we can reason in terms of scientific progress. There is a separate set of questions about philosophical progress, including notably, the question of whether there are results of a kind that philosophy aims for and science does not, philosophical but not scientific progress. But we need not answer these questions to address the debate about the comparison between the success of science and philosophy; reasoning in terms of scientific progress in philosophy can suffice.

Second, I propose that instead of choosing one given definition of scientific progress, one can evaluate the success of philosophy by studying whether philosophy displays the markers associated with all definitions, regardless of what definition is correct. In articulating this proposal, I consider a variety of possible markers of scientific progress, expanding the catalogue that Dellsén et al., (2022) have offered (knowledge, verisimilitude, problem-solving, understanding, justification, knowledge-how, pragmatic benefits, and scientists’ subjective or intersubjective goals).

This approach has costs and benefits. The main benefit is that it allows us to get more stable results. A view about scientific progress in philosophy that presupposes a given definition of scientific progress can only be as good as the definition. Suppose, for instance, that we endorse a view such that scientific progress is associated with the accumulation of knowledge. There is much debate about this view (Bird, 2007, 2016; Cevolani & Tambolo, 2013; Niiniluoto, 2014; Dellsén, 2016; 2017; 2021; Rowbottom, 2014; Dellsén et al., 2023). Therefore, our view about scientific progress in philosophy is quite unstable: if a new argument emerges, it may very well be undermined. On the other hand, if we evaluate how much scientific progress philosophy makes by the standards of all the accounts, then all that remains to do is discuss which account is correct, if any, but this debate could not undermine our results.

The main cost of this approach is that it is more demanding in terms of the amount of empirical research it requires. There are several accounts of scientific progress, and in many of them, scientific progress or lack thereof can be detected by considering different markers. For instance, in some accounts, persistent peer disagreement indicates that scientific progress is not occurring while in other accounts it does not (Dellsén et al., 2023). My approach requires considering all the markers that have been associated with scientific progress. Thus, it is more demanding than choosing one account and focusing on the markers relevant to that account.

Having conceded that there are both costs and benefits, I argue that the benefits outweigh the costs. The amount of empirical research required to consider all possible markers of scientific progress is more manageable than it might initially seem. Admittedly, the approach I propose requires more empirical research compared with Dellsén, Lawler, and Norton’s (2022) approach. But this cost is not great enough to make my proposal non-viable, considering its benefits. I illustrate this by discussing concretely how my proposed approach can be used to assess the success of philosophy relative to science.

The rest of this paper has the following structure. In section 1, I present a framework for the debate, and I make a few preliminary remarks to clarify the terminology I use. In section 2, I give an overview of several definitions of scientific progress. In section 3, I condense them into a single list of markers for scientific progress and articulate my approach. In section 4, I discuss what possible results we could get using this list to analyse the success of philosophy, and I articulate the main advantages of this approach. Conclusions follow suit.

1 Philosophical progress and scientific progress

We may distinguish two questions about progress in philosophy.

Exceptionalism question

How much scientific progress does philosophy make, compared with science?¹

Inner question

How much philosophical progress does philosophy make?

Throughout the article, I use scientific progress to refer to results of the kind that science aims to achieve, whatever these may be, and I use philosophical progress to refer to achievements of the kind that philosophy aims to achieve, whatever these may be. If philosophy aims at achievements of the kind that science aims to, then at least some philosophical progress is scientific progress. If philosophy aims only at achievements of the kind that science aims to, then all philosophical progress is scientific progress. If philosophy aims at some achievements of a kind that science does not, then some philosophical progress is not scientific.² Given this, the exceptionalism question is connected to the debate on whether philosophy is exceptional: to what extent does philosophy achieve the results that science achieves?³ By contrast, the inner question tries to evaluate philosophy by its own standards: does philosophy achieve the results it tries to achieve, which may or may not be the same sort of results that science aims to achieve?

It is important to distinguish these two questions because one can give quite different answers to each of them. If philosophical progress and scientific progress are the same, then the two questions coincide, but one can deny that they are. Wittgenstein, for instance, held that “philosophy is not one of the natural sciences” because, unlike science, “philosophy aims at the logical clarification of thoughts” (1921, §4.111, §4.112). Thus, for Wittgenstein, philosophy can make philosophical progress through the clarification of thought even though it does not make any scientific progress. Similarly, for Beebee (2018), philosophy accumulates no knowledge, which is the principal aim of science, thus it makes no scientific progress, but philosophy can aim at reflective equilibrium to make philosophical progress.

Kant and Russell, in the passages I quoted above, were addressing the exceptionalism question. That is, they express views about the extent to which philosophers have obtained scientific results, drawing a comparison with science. Similarly, the views I referred to defended by Dietrich (2011), Beebee (2018), Chalmers (2015), Wilson (2017), Stoljar (2017), and Cappelen (2017) are all views about the exceptionalism question. Dellsén, Lawler, and Norton intend to provide a framework for this debate, but then they concentrate on philosophical progress on the assumption that philosophy and science “have at least some aims in common” (2022, 819).⁴ Indeed, the definitions they offer are all inspired by accounts of scientific progress, but they are explicitly formulated as views about what philosophical progress amounts to. This may create some confusion by opening the door, as Dellsén, Lawler, and Norton themselves put it, “for optimists to celebrate achievements that pessimists do not deny, while pessimists decry limitations that optimists are happy to accept” (2022, p.815). For instance, Beebee (2018) can continue to argue that philosophy makes no scientific progress because it cannot accumulate knowledge, without denying that it makes philosophical progress, reaching reflective equilibrium, by the standard of one of Dellsén, Lawler, and Norton’s definitions.

In order to avoid this sort of confusion, it is best to reason about the exceptionalism question and the inner question separately. That is, it is best to discuss separately to what extent philosophy reaches science’s aims, and to what extent it reaches its own aims.⁵ The debate with which I started focuses on the exceptionalism question. Accordingly, in the rest of this paper, I will consider the exceptionalism question only. I remain neutral on the inner question for the time being. That is, the methodology I propose presupposes no view about the nature and prevalence of philosophical progress; those are issues that will have to be addressed at another juncture.⁶

Now, having framed the debate I focus on, there are a few clarificatory remarks to make about the terminology used in this debate. Some participants of this debate use the term optimism and cognates to refer to views in this debate (Chalmers, 2015; Stoljar, 2017; Dellsén et al., 2023). Schematically, the range of possible views can be summarised as follows:

Optimism

Philosophy has made a great deal of scientific progress, possibly as much as the most successful sciences, or more.

Moderate optimism

Philosophy has made a comparatively high amount of scientific progress.

Moderate pessimism

Philosophy has made a comparatively low amount of scientific progress

Pessimism

Philosophy has made a great deal less scientific progress than the sciences, possibly none.

This terminology is quite misleading because ‘optimism’ and ‘pessimism’ are bias-inducing terms. To appreciate this, notice that if no science makes any scientific progress, then trivially optimism is true; philosophy makes at least as much scientific progress as the most successful scientific disciplines do. One can argue that there is nothing wrong with science if this view is true (Rowbottom, 2014; 2023), but it seems fair to assume that most would not take this view to be optimistic. Indeed, in a sense, pessimism and moderate pessimism are the more optimistic views—assuming that scientific progress is desirable—because, unlike optimism and moderate optimism, they entail that science makes scientific progress.⁷ To avoid confusion, it would be best to avoid using bias-inducing terminology to label philosophical views. However, ever since Chalmers (2015) introduced these terms, it has become customary to use them in the metaphilosophical debate. Thus, attempting to introduce a new non-bias-inducing terminology—say by replacing optimism with anti-exceptionalism and pessimism with exceptionalism—could result in further confusion. Thus, I opt for a compromise. In the rest of this paper, I use the familiar terminology having explicitly clarified that the less scientific progress there is in science, the lower the bar is to argue that philosophy is not exceptional and hence optimism is true.

A final warning to make is that one must not apply a double standard. The exceptionalism question is about the rate of scientific progress in philosophy, compared with the rate in science. In order to make this comparison, one must evaluate scientific progress in philosophy by the same standard that one uses to evaluate scientific progress in science. Otherwise, one would end up measuring different markers in science and philosophy, thus undermining the comparison.

This warning might seem trivial, but it is worth making it because in some cases double standards can be difficult to detect. Take Chalmers’s (2015) defence of moderate pessimism, for instance. A close inspection of Chalmers’s argument reveals an instance of a double standard that could easily pass unnoticed. Chalmers suggests that scientific progress in mathematics should be evaluated based on its success in answering questions posed at the turn of the 20^th century, such as the Hilbert (1902) problems. Given this, applying a single standard would require Chalmers to also evaluate scientific progress in philosophy based on its success in answering questions posed by philosophers at the turn of the 20^th century. However, one of the questions that Chalmers considers is the following: is logic classical or non-classical? At the turn of the 20^th century, non-classical logics had not yet been developed to the level of sophistication that classical logic had; Łukasiewicz’s (1910) foundational work about non-classical logics—among others—would only become widely available later. Indeed, Russell, in the work that Chalmers quotes, The Problems of Philosophy (Russell, 1912), does not consider questions about logical system choice at all. It is important to keep in mind the point that we should not apply double standards, so that these sorts of cases do not go unnoticed.

2 Views about scientific progress

Several different definitions of scientific progress have been offered in the literature. In this section, I present the eight views that are currently live options in the debate regarding scientific progress in the philosophy of science. I begin by briefly presenting the four views that Dellsén et al., (2022) have already reviewed, and then I consider the other four.

Epistemic view

Scientific Progress occurs if and only if new scientific knowledge is discovered.

This is the view that most philosophers throughout history, including Kant, Russell, and Wittgenstein, endorsed.⁸ Most recently, after a brief period in the second half of the 20^th century during which several philosophers of science rejected it, it has been advocated by Bird (2007, 2008, 2016).⁹

The epistemic view is not monolithic because what it contends exactly depends on the underlying account of knowledge, which is itself a matter of much debate (Borges et al., 2017). Bird (2007) builds his view on Williamson’s (2000) account of knowledge, but different accounts could be used. One could argue that this opens a Pandora’s box: there should be an entirely different discussion cataloguing possible definitions of knowledge before moving on to the next view. But this can be avoided by observing that the pre-theoretical notion that various accounts of knowledge are trying to capture is the same, or at least there are very considerable overlaps. Indeed, virtually anyone would agree that belief, truth, and justification are necessary conditions for knowledge. In the interest of keeping the discussion manageable, we can focus on the common pre-theoretical notion instead of considering any specific definition of knowledge.¹⁰

Semantic view

Scientific progress occurs if and only if the corpus of scientific beliefs becomes more verisimilar.

The semantic view is usually associated with Popper (1934), who set it forth as an alternative to the epistemic view in his broader anti-inductivist project. Recently, this view has been defended by Cevolani and Tambolo (2013) and Niiniluoto (2014).

The central notion here is that approximation to the truth is a spectrum, a corpus of beliefs can be more or less verisimilar. This can be understood either as the view that the individual beliefs can be more or less verisimilar or as the claim that the verisimilitude of a cluster of beliefs is a function of how many true beliefs there are in the cluster, but individual beliefs are either true or false. Several formal accounts have been developed to try and make this more precise (Niiniluoto, 1998), but there is no need to review them here. For the present purposes, I only need to make two clarifications about the notion of verisimilitude. First, the verisimilitude of a corpus of beliefs can increase not only when true or verisimilar beliefs are added but also when false or non-verisimilar beliefs are removed.¹¹ Second, the epistemic view needs to also employ the notion of verisimilitude, or perhaps some other notion that plays the same theoretical role, if it is to talk about knowledge about theories because any scientific theory contains some falsehoods (Bird, 2007, p.76–78).¹² In this case, scientific progress in the semantic account is a necessary but not sufficient condition for scientific progress in the epistemic account.

Functionalist view

Scientific progress occurs if and only if scientists solve or formulate a problem.

This view is usually associated with Kuhn (1962), who set it forth as an alternative to both the epistemic and the semantic view in his broader anti-realist project. This view was later developed by Laudan (1977; 1981a) and it has more recently been defended by Shan (2019).

Given its association with Kuhn’s anti-realism, this view is sometimes described as internalist: “whether or not a scientific community has made [scientific] progress […] does not depend on features that may be unknown” (Bird, 2007, p.69). Now, it may be true that Kuhn endorsed internalism—although one can also argue that he did not (Rowbottom, 2023, p.27–30)—but, more generally, the functionalist view does not entail internalism. One can hold that it depends entirely on scientists’ known opinion whether they have solved a problem or created a new problem, in which case this view is internalist. But one can also introduce other factors that may be unknown to determine whether a solution to a problem has been found or a new problem has been formulated. Indeed, quite in general, any view about scientific progress can be made internalist or externalist. The epistemic view, for instance, is internalist if the underlying definition of knowledge is.

Noetic view

Scientific progress occurs if and only if there is an increase in scientific understanding.

The idea that scientific understanding is crucial to scientific progress has a long history, having been endorsed by the likes of Aristotle. However, Aristotle held that understanding requires knowledge.¹³ Thus, his views about the value of understanding do not constitute a rejection of the epistemic view; at most, they might be a restriction of the epistemic view: scientific progress occurs only when a certain type of knowledge, namely understanding, is discovered. A more contentious version of the noetic view is the view that understanding, instead of knowledge, is the aim of science. Finocchiaro (1976) suggests that understanding—roughly understood in terms of the intelligibility of theories used in solving scientific problems—may be central to progress. More recently—in the wake of a renewed interest in the notion of understanding among epistemologists and philosophers of science (Grimm, 2024)—Dellsén (2016; 2017; 2021) developed a view in which there can be understanding and thus scientific progress without knowledge or justification; in this view understanding is conceived roughly in terms of correctly explaining or making predictions about phenomena.¹⁴

The contention of the noetic view differs depending on the underlying notion of understanding. In a nutshell, if understanding is a type of knowledge or knowledge is a necessary condition for understanding, then the noetic view is a restriction of the epistemic view. That is, scientific progress in the epistemic account is a necessary but not sufficient condition for scientific progress in the noetic account. If understanding is not a type of knowledge and is not a necessary condition for it—say, because understanding is not factive or it does not require justification—then there can be epistemic progress without noetic progress and vice versa.

Let us turn now to the four views which are not discussed by Dellsén et al., (2022).

Justification view

Scientific progress occurs if and only if there is an increase in scientific justification.

Justification is a necessary condition for knowledge. Thus, any advocate of the epistemic view would recognise the importance of justification. However, virtually everybody who works on scientific progress denies that justification without truth or verisimilitude is sufficient for scientific progress. Stegenga (2024) is the exception to this, in his view, truth is “a convenient benediction” (553), but it is not necessary.

The exact content of the justification view differs depending on the underlying notion of justification. It is agreed that justification is not factive: one can be highly justified in holding a false belief. It is less clear, on the other hand, whether the justification of any given belief should be understood as a spectrum or rather justification is a binary notion and increases in justification correspond to increases in the ratio of justified beliefs over unjustified ones (Hawthorne & Logins, 2020). Stegenga (2024) argues for the former view and articulates an account of scientific progress through increases in justification of individual beliefs based on Sprenger and Hartmann’s (2019) work on Bayesian confirmation theory. But other accounts are possible. As usual, I avoid opening a Pandora’s box here.

Procedural view

Scientific progress occurs if there is an increase in scientific know-how or new tools and methods are developed for scientists to use.

There is a long tradition associating scientific progress with the development of new methods and procedures. Bacon and Galileo, for instance, sought to promote scientific progress by developing scientific methods. However, new methods and procedures have traditionally been seen as means to an end rather than an end in themselves. Recently, Mizrahi (2013) argued that scientific progress can occur when scientists develop a new tool or a new procedure even if no further cognitive advance is made. Similarly, Rowbottom argues that scientific progress can occur when scientific understanding expands through the development of conceptual tools that are “cognitively appropriate to serving scientists’ empirical ends” (2019, p.3).

One can in principle develop a view in which an increase in know-how is also a necessary condition for scientific progress, but this has not been done. That is, advocates of the procedural view accept that there can be scientific progress without new know-how or tools. If knowledge-how is a type of knowledge-that, as Stanley and Williamson (2001) have argued, then this view states that a particular type of scientific progress in the epistemic account is a sufficient condition for scientific progress. Otherwise, one can endorse this view while rejecting that the discovery of any sort of knowledge-that is sufficient for scientific progress.

Pragmatist view

Scientific progress occurs if and only if science achieves results that are significant for possibly parochial human needs.

Pragmatists argue that scientific progress should not be understood as a teleological progression towards an ideal aim, say, true theories, but rather in terms of solving problems that are significant for some people at a given time. The discovery of penicillin, for instance, solved a significant problem in that it met some human needs, although this was parochial since some humans need penicillin more badly than others and some of those who need it badly have no access to it. Kitcher (2015) recently articulated a version of this.

Just as the functionalist view, the pragmatist view tends to be associated with scientific anti-realism, and as such, one might assume that it is internalist. But, as all the other views do, the pragmatist view leaves the internalism-externalism question open. One can argue that it is entirely up to factors known to scientists—or some other part of society—whether a given problem is significant, but the pragmatist view is not automatically committed to this. That is, there might be factors that need not be known that make a problem significant. Moreover, one can avoid internalism by introducing externalist criteria to assess whether science has achieved a result. Kitcher (2015) holds that significant results must constitute true answers to significant questions.

Deflationism

There is no necessary or sufficient condition for scientific progress, our claims about scientific progress reflect subjective or intersubjective value judgements.

In many philosophical debates, there are some authors arguing that the debate should be discontinued because the problem it tries to solve is spurious. Deflationism fulfils this role in the debate about scientific progress: we cannot agree on what the necessary and sufficient conditions for scientific progress are because there are none. This view has been most recently defended by Rowbottom (2014; 2023).

Deflationism acknowledges that we do talk in terms of scientific progress and that this can be useful to express subjective or intersubjective value judgements about the success of science. However, attempting to find objective criteria for this is akin to trying to decide “what’s the best flavour of ice cream?” (Rowbottom, 2023, p.31). It is in principle possible to articulate a version of deflationism in which scientists lack even subjective or intersubjective aims. This has not been defended, though, because it contradicts the obvious fact that scientists’ claims about the successes of their discipline must reflect at least some psychological or social facts.

Like the functional and pragmatist views, this view is naturally associated with a broadly anti-realist position—Rowbottom (2019) himself is an anti-realist—and as such one might assume that it is internalist. But in fact, the view that claims about scientific progress reflect subjective or intersubjective value judgement does not entail that scientific progress depends on facts that must be known. One can argue that people are not always aware of their subjective preferences or of the intersubjective value judgements that their society has adopted; thus, scientific progress depends on factors that may be unknown.

In conclusion, I should mention that one can also take a pluralist stance: more than one of the conditions I discussed are individually sufficient, and there is no other sufficient condition for scientific progress. For instance, one can endorse a semantic-procedural pluralism in which scientific progress occurs if and only if there is an increase in verisimilitude or knowledge-how. A case in point for this is Mizrahi (2013), who argues that know-how is a sufficient but not necessary condition for scientific progress, thus allowing for more sufficient conditions.¹⁵

Making explicit what possible pluralist positions there are is slightly complicated because the relations between the views are not obvious. If understanding is a type of knowledge, then a noetic-epistemic pluralism is just the epistemic view since epistemic progress is a necessary condition for noetic progress. Otherwise, there can be a noetic-epistemic pluralism distinct from the epistemic view. Similarly, if increases in verisimilitude are a necessary condition for epistemic progress, then a semantic-epistemic pluralism is just the semantic view. Examples can be easily multiplied, but there seems to be no need to give a full list of the possible pluralist positions.

3 An approach to search for scientific progress in philosophy

Dellsén et al., (2022) argue, in effect, that given that there is no consensus about what definition of scientific progress is correct, views about the exceptionalism question must incorporate views about scientific progress.¹⁶ Thus, for instance, Beebee’s (2018) view would be the following.

Epistemic pessimism

Scientific progress amounts to the accumulation of scientific knowledge and philosophy accumulated a great deal less scientific knowledge than science.

In this approach, views about the definition of scientific progress do a great deal of theoretical heavy lifting. If views about the exceptionalism question incorporate views about the definition of scientific progress, then they can be challenged by rejecting the definition of scientific progress without challenging any factual claim about philosophy. For instance, the following view—which Dellsén et al., (2024) sympathise with—rejects epistemic pessimism on the sole ground that the epistemic view is false without rejecting any factual claim that epistemic pessimism makes about philosophy.

Noetic moderate Optimism

Scientific progress amounts to the accumulation of scientific understanding and philosophy accumulated a comparatively high amount of scientific understanding.

Questions about scientific progress in philosophy can boil down to a disagreement about the definition of scientific progress just as, say, discussions about whether a given politician is a fascist can focus only on the definition of fascism without discussing empirical questions about the politician.

I do not dispute that this is one possible approach. Clearly, it is possible to discuss the exceptionalism question by focusing on views such as epistemic pessimism and noetic moderate optimism. What I want to do is articulate another approach and discuss some of its merits. My proposal, at its core, is that we can assess questions about the success of philosophy compared with science independently of what definition of scientific progress is correct. We can assess whether philosophy accumulates knowledge, understanding, know-how, and so forth regardless of which is associated with scientific progress, and by doing this, we can already say a great deal about the exceptionalism question.

This is the full list of markers associated with scientific progress that emerges from considering the debate:

Possible markers of scientific progress

• (a) The corpus of scientific knowledge expands

• (b) The verisimilitude of scientific theories increases

• (c) Scientific problems are solved

• (d) New scientific problems are formulated

• (e) The corpus of scientific understanding expands

• (f) The corpus of scientific justification expands

• (g) The corpus of scientific know-how expands

• (h) New tools for scientific enquiry are developed

• (i) Scientific results are achieved meeting possibly parochial human needs

• (j) Scientists achieve their subjective aims

• (k) Scientists achieve their intersubjective aims

I should make two clarifications about this. First, the list is neutral about controversial relationships that there may be between the various conditions. If understanding is a type of knowledge, for instance, then (a) is always satisfied when (e) is. My proposal is to remain neutral on these issues. Dependence relations between the items of the list should be accepted only when they are uncontroversial, such as in the case of knowledge (a) entailing justification (f). Second, the list could be expanded by noticing that each of the conditions can be doubled into an internalist and an externalist condition. But again, this would make the discussion unnecessarily fine-grained. We can discuss whether philosophy meets each of these conditions setting aside for the moment whether the conditions should be understood as externalist or internalist.¹⁷

Now, my proposal is that we can discuss the exceptionalism question by evaluating which conditions in this list, if any, are met by philosophy, and to what extent without choosing one condition. Is this proposal viable? Is it possible, and feasible, to evaluate scientific progress by the standards of all existing accounts?

A straightforward answer is that if questions about each account are tractable, then questions about all accounts are also tractable. If it turned out that one of the accounts makes empirical questions about scientific progress intractable, that would be a reason to discard it. But this would not undermine the claim that we can evaluate scientific progress in philosophy by the standards of all accounts, it would only remove one account from the list.¹⁸

This answer is somewhat uninformative since it gives us no information about how concretely to evaluate scientific progress by the standards of each view. My more informative answer is that we should use historical case studies comparing major turning points in the history of philosophy, such as, say, the development of predicate calculus, with major turning points in the history of science, such as, say, the development of general relativity. These case studies should consider each of the markers of scientific progress to evaluate how philosophy compares with science.

Why use case studies? We should not expect to be able to evaluate scientific progress in philosophy in terms that can be straightforwardly quantified. Bourget and Chalmers (2014) attempted to use a quantitative survey to measure levels of consensus in philosophy, which, according to Chalmers (2015), are correlated with convergence to the truth.¹⁹ There are some problems with this (Capellen 2017; Dellsén et al., 2023), but even if we set this aside, clearly the survey gives us no information about changes over time. Indeed, Chalmers (2015) himself admits that the survey returns a partial picture.

For fuller data […] we would need the results of the PhilPapers Survey not just in 2009, but at regular intervals in the past: 1909, 1809, and so on. At each point we would need to ask members of the philosophical community first, what they take to be the big questions of philosophy and second, what they take to be the answers to those questions as well as to big questions from past surveys. We would also need to have analogous longitudinal surveys in other fields. (Chalmers, 2015, p.10).

In other words: evaluating scientific progress in science and philosophy requires the use of historical methods, which are unlikely to give straightforward quantitative results when it comes to measuring phenomena such as convergence to the truth or expansion of understanding. It seems unlikely that we will ever be able to say to a reasonable degree of approximation that, say, 70 big questions in physics were solved in the Middle Ages while 8 big questions in philosophy were solved. Identifying questions that medieval thinkers were working on and assessing if they answered them is bound to involve qualitative considerations which are unlikely to return straightforward quantitative measurements. So, what can we do? How can we assess to what extent philosophy has any of the markers in the list above, given the limitations that the historical nature of the exceptionalism question imposes?

When we look at historical accounts of a scientific discipline, typically, we find extensive accounts of the work of scientists towards an achievement that is recognised as a major milestone in the history of the discipline. Examples of this include, for instance, accounts of the development of general relativity. Because the history of science escapes straightforward quantification, measuring exactly how much physicists approached the truth between 1881—the year of the first of the Michelson and Morley experiments—and 1916—when Einstein (1916) consolidated his various results in a single article about General Relativity—is very difficult. Nevertheless, historians have qualitatively analysed that period with a special emphasis on some of the most significant findings.

Historical methods prevent us from giving precise measurements, but analyses of major collective achievements in science can give us strong indicators of whether some conditions for scientific progress have been met. We can apply the same approach to comparatively analyse the history of philosophy. It seems difficult to assess exactly to what extent philosophers approached the truth between 1893—when Frege published his Grundgesetze—and 1913—when Whitehead and Russell published the third volume of Principia Mathematica—but we can have a case study to assess whether the development of predicate calculus constitutes a major achievement comparable with the development of general relativity.

Historical studies into convergence to the truth or expansion of understanding are unlikely to lead to straightforward quantitative results. But we can do case studies about major milestones in the history of philosophy in which the discipline made a recognised leap forward and compare them with milestones achieved by scientific disciplines for each of the markers above. If, for any given marker, philosophy is on a par with the most successful scientific disciplines, then plausibly optimism is true relative to that marker. Otherwise, depending on how many milestones we find and how comparatively significant they are, moderate optimism or some version of pessimism may be true. This method does not allow us to make a very fine-grained comparison in the way that a quantitative measurement would, but again, the historical nature of the exceptionalism question prevents straightforward quantitative measurements. Comparing major intellectual turning points in the history of science and philosophy gives us a feasible way to approach a problem that may seem intractable if we fixate on an idealised level of measurement precision.

Let us consider a few objections now, addressing which can help articulate the proposal further. In the recent debate on the methodology of philosophy of science, several authors—including, among others, Pitt (2001), Kinzel (2015; 2016), and Chakravartty (2017)—have raised concerns with the well-established practice of using historical case studies. One could reasonably worry that these issues may also undermine the methodology I am setting forth.

A straightforward answer to this worry is that, as I argued, the exceptionalism question requires qualitative historical methods. There is no way, as Chalmers puts it, to do a “PhilPapers Survey not just in 2009, but at regular intervals in the past: 1909, 1809, and so on” (2015, p.10), and more generally, it seems unlikely that might be able to quantitatively ascertain to what extent any given marker of scientific progress is present in the history of philosophy, especially when we go far back. Even though qualitative historical studies have limitations, as any philosophical method does, we need to use them if we are to address the exceptionalism question at all. We could abandon the exceptionalism question altogether, but then by the same reasoning we would have to abandon any question addressing which requires qualitative historical methods, and that seems unreasonable.

This answer is somewhat uninformative since it gives us no information about how we may try to address the concerns that have been raised with case studies. I cannot discuss the details of this fully without opening a Pandora’s box, but I can give a more informative answer drawing from Bolinska and Martin’s (2020) analysis of the debate about case studies in the history of science.

Bolinska and Martin (2020) distinguish two sorts of worries with case studies. First, we have methodological worries having to do with the possibility of biases in the selection, construction, or analysis of a case study. Second, we have what they call metaphysical worries having to do with the impossibility of drawing necessary, or even just general, conclusions from contingent episodes in the history of science.

Now, as Bolinska and Martin (2020) convincingly argue, the former sort of worry does not seem to pose a challenge to the use of case studies beyond the familiar point that in general there may be biases in the selection, construction, and analysis of a piece of evidence. Experiments and philosophical arguments, for instance, can also be selected, constructed, and analysed in a biased way. Thus, the first sort of objection only indicates that the familiar precautions against biases should also be in place when we use historical case studies. As for the second objection, Bolinska and Martin (2020) develop an account of what they call canonical case studies to show that general conclusions can be drawn from at least some case studies. But endorsing this solution does not seem necessary when it comes to case studies attempting to address the exceptionalism question. Even if case studies cannot provide general conclusions, they can provide meaningful insights into the exceptionalism question. Indeed, the claim that there was progress, or that there was no progress, at particular junctures in history in which science was at a major turning point could go a long way towards assessing the success of philosophy relative to science.

Another objection that I tend to get when I present this is that this approach makes contentious assumptions about the history of science. As the debates about the pessimistic meta-induction and the unconceived alternatives argument indicate, it is not at all obvious to what extent science has converged towards the truth in major scientific turning points. Depending on what philosophical arguments one endorses, one can get to radically different views (Laudan, 1981b; Psillos, 1999; Lyons, 2006; Stanford, 2006). We should expect similar issues to arise in attempting to measure convergence towards several of the aims in the list above. Thus, it is not a neutral view to assume that science has met at least some of the conditions associated with scientific progress.

This problem can be resolved by observing that, as I pointed out above, the view that science makes no scientific progress entails optimism. Counterintuitively enough, if science is not successful at all then optimism—the view that philosophy is at least as successful as science—is trivially true. This is the case no matter what condition is associated with scientific progress. Thus, for each given condition for scientific progress, the less successful science is, the lower the bar is for philosophy to be as successful as science. If science has failed to converge towards the truth, then optimism is trivially true with respect to the verisimilitude condition (b). If science has failed to expand understanding, then optimism is trivially true with respect to the understanding condition (e), and so forth. In order to even consider that optimism may be false, we need to assume that science has been at least moderately successful. The point is not that science does in fact accumulate truths, knowledge, understanding, and so forth, but rather that if it does not, then at least some forms of pessimism are non-starters, and more generally the least successful science is with respect to any marker, the less problematic the exceptionalist question becomes, relative to that marker.

The other objections I tend to receive when I present this proposal concern its practical feasibility. One can argue that my proposal is practically unfeasible because the list of conditions is too long. In order to pursue my proposal, one would need to develop detailed case studies comparing major turning points in philosophy with major turning points in science for each of the markers of scientific progress. This would require so much work that, although this is in principle possible, it is not in practice feasible.

Now, it is true that my proposal involves more empirical research than Dellsén, Lawler, and Norton’s (2022) approach. If we first choose one account of scientific progress and then measure how much scientific progress philosophy has made, we can avoid some empirical work thanks to the theoretical heavy lifting that the argument for our favourite account is doing. This being said, we should not overestimate the amount of additional empirical work that my proposal involves.

We have reason to believe that at every major scientific turning point—again, unless we endorse a view in which optimism is straightforwardly true relative to some marker—there was scientific progress not just by the standards of one account, but by the standards of all accounts. Take the development of special relativity, for instance. At this juncture, unless one is very impressed by arguments such as the pessimist meta-induction, one should agree that knowledge was produced—physicists have knowledge about the results of the Michelson and Morley experiments, among other things—which indicates expansions of justification and verisimilitude too. Problems with the theory of the luminiferous aether, among others, were solved and novel problems related to generalising special relativity to gravitational phenomena emerged. Understanding of the relationship between mass, energy, and the speed of light was expanded. New tools were developed—including, for instance, the experimental methodology that Michelson and Morley used—and scientists accumulated know-how on how to use these tools. Moreover, this development served to satisfy some human needs—it serves as the foundation for GPS technology, for instance—and as most of the things that people celebrate, it satisfied the subjective and intersubjective aims of some of the people involved. No matter how one looks at it, the development of special relativity constitutes scientific progress.²⁰

If any of the accounts of scientific progress focused on a marker that clearly is not present at a major turning point in science, that would be a straightforward reason to discard it. If, for instance, there plausibly had been no expansion of understanding during the Copernican revolution, that would be a straightforward reason to discard the noetic view since there was scientific progress during the Copernican revolution if there ever was. In this scenario, opponents of the noetic view would simply point out that scientific understanding was not expanded during the Copernican revolution to make the point that the view is extremely implausible. But this is not what we see in the debate. All accounts have detractors, but nobody argues that some accounts can be straightforwardly discarded in this way.

Opponents of the noetic view may argue that understanding is really a type of knowledge (Bird, 2007), but crucially, they do not argue that there was no expansion of understanding during major turning points in the history of science. This seems to indicate that even for the detractors of the noetic view, understanding is correlated with progress and expansions of understanding are present at major turning points of science. Analogously, Dellsén (2016) argues that in his article about Brownian motion, Einstein (1905/1956) expanded understanding but not knowledge and therefore—since the article made scientific progress—scientific progress should not be associated with knowledge. However, and crucially, Dellsén would not deny that over the period in which general relativity was developed, much new knowledge was acquired by the community of physicists and indeed physicists in general accumulated knowledge. Stegenga (2024) argues that truth is not necessary for scientific progress, but he acknowledges that scientists have discovered a great many truths, especially in connection to major scientific turning points, and similarly, nobody seems to deny that at major scientific turning points new problems were formulated and know-how expanded. Even Popper conceded that what Kuhn calls “normal science” is widespread, although he insists that it does not constitute progress (1970, 52).²¹

In general—again, unless they endorse views in which optimism is straightforwardly true relative to some markers—philosophers of science working on scientific progress accept that all the achievements in the list above are correlated with scientific progress and that in particular at major turning points in the history of science, all the markers are present. That is, the debate concentrates on limit cases in which one marker but not another is present to try and assess what achievements are not just correlated with scientific progress but necessary and sufficient for it.²² This consensus gives us strong reasons to think that all the markers listed above are indeed correlated with progress; if a plausible case could be made that some marker is absent during a major turning point, we would expect a detractor of the relevant account to have pointed this out.

The observation that at major scientific turning points all markers of scientific progress are present removes the worry that my proposal is significantly less feasible than Dellsén et al., (2022) proposal. We need not have one case study for verisimilitude, one for understanding, one for know-how, and so forth. Science has met all the conditions for scientific progress during the periods in which it is recognised to have made major leaps forward. If philosophy is not exceptional, then we should expect it to also have all the markers of scientific progress in analogous periods.

Another worry concerning feasibility is that my proposal is unfeasible, even though it does not require significantly more empirical work than Dellsén et al., (2022). One could argue, in particular, that even if we concentrate on a relatively short period—say, the period between the late 19^th century and early 20^th century—we would need a myriad of case studies to meaningfully tackle the exceptionalism question. Finding out, say, that the development of predicate calculus constitutes progress would not be enough, since at best it would indicate that philosophy of logic and philosophy of language made progress over the relevant period. We would also need case studies about developments in other philosophical fields of inquiry. Indeed, to establish to what extent there was scientific progress in philosophy compared to science, over the relevant period, we would need to develop one case study for each of the developments that occurred in the more successful scientific disciplines over the same period.

Considering this objection gives me a chance to illustrate another reason why I have been focusing on major turning points in the history of science and philosophy, rather than taking a more fine-grained approach focusing, say, on works done by individual authors on a given year. A major turning point in the history of science, or philosophy, may begin within the confines of a relatively narrow subfield—the discovery of oxygen, for instance, begun with relatively specific questions about combustion—but over time, it leads to so many new findings with such broad implications that it has an impact on the entirety of its field. The development of predicate calculus began within the confines of philosophy of logic and philosophy of language, but had such far-reaching implications that it is hard to think of any area of contemporary analytic philosophy that is unaffected by it. Considering cases with such far-reaching implications avoids the risk of reaching conclusions that only apply to a given subfield.

As for the worry that my proposal—or indeed Dellsén, Lawler, and Norton’s (2022) proposal—requires us to develop a case study for each development that there was in successful scientific disciplines over the same period, first of all, again, we must not apply a double standard. It would not be reasonable to expect that we may find a major turning point in the history of philosophy for every minor finding that was made in science over the same period. On both sides, we must concentrate on milestones that clearly represent a major step forward for the discipline. If we do that, though, the suggestion that we may need a myriad of case studies for every historical period loses much of its substance. Mature sciences do not have extraordinary peaks of activity every decade. There may be periods in which a given science is extraordinarily successful, but these are the exception rather than the norm, and even in such cases, historians can usually point out relatively few major turning points. Physics, for instance, underwent a period of extraordinary success in the period between the late 19^th century and the early 20^th century, but historians cannot point out dozens of turning points as significant as the development of relativity theory. Philosophy should be evaluated by the same standard, and as such, the analysis of relatively few major developments—or indeed even one major development—can give us a great deal of information about any given period.²³

Finally, one could object that the methodological approach I am setting forward is too coarse-grained. Considering in a single case study processes as complex as the development of predicate calculus, this objection goes, can at best return a distorted picture. To respond to this, I can only observe that the exceptionalism question is coarse-grained. We ask: does philosophy, a discipline with thousands of years of history and just as many subdisciplines, make scientific progress? This question invites us to look at the bigger picture, although it is true that one should not lose sight of the trees when looking at the forest. We can, of course, look with a much more fine-grained approach at several narrower developments in the history of philosophy—that can be very profitable for other purposes—but it is unlikely to help us address the question we began with. There is a real risk that if we insist on ever more very fine-grained analyses, we only fuel the existing impressions that philosophy pursues an idealisation of rigour and precision to the point of never beginning to address its questions.

4 Possible results and the advantages of this approach

We can divide into two groups the possible outcomes of a case study looking into a major turning point in the history of philosophy and assessing whether, at that juncture, philosophy has the markers associated with scientific progress.

Clear case for optimism

Over a given period, philosophy is on a par with the most successful sciences of the time with respect to all markers.

Unclear case for pessimism

Over a given period, philosophy underperformed with respect to some but not all markers.

The option that philosophy underperformed with respect to all markers seems a non-starter since it seems quite trivial that it has some of the markers. It seems hard to question, for instance, that philosophers over history have met some of their subjective aims (j) and they have developed some know-how on how to debate the philosophical issues of their time (g).

What can each of the possible results tell us about the exceptionalism question? If we have a clear case for optimism, then the case supports optimism, at that point in history, if any of the accounts discussed above are correct. In order to fully ascertain whether optimism is true, it would be necessary to consider more case studies, but each clear case for optimism supports optimism. If any of the proposed accounts is correct, philosophy at that juncture made scientific progress by the standards of the correct account, since the case meets the standards of all proposed accounts. In other words, this sort of result allows us to circumvent the debate about the analysis of scientific progress, assuming that one account is correct.²⁴

What if, on the other hand, philosophy at a given juncture underperformed significantly with respect to some markers? Suppose, for instance, that in the period in which Frege and Russell were working on logicism, their celebrated theories of reference, and predicate calculus (circa 1893 to 1913), philosophy underperformed only with respect to justification (f) and knowledge (a). What would this tell us about the exceptionalism question?

As I observed above, unless we take a view in which optimism is quite straightforwardly true, relative to some markers, we can assume that all the markers associated with scientific progress are correlated with it. Dellsén (2016) denies that scientific progress coincides with the accumulation of knowledge, but he accepts that scientists accumulated a great deal of knowledge. Therefore, irrespective of the debate about scientific progress, if we were to find that over a given period philosophers have failed to accumulate justification and knowledge, that would indicate that philosophy, at that juncture, is exceptional with respect to its results. Philosophy differs from science—to a degree that depends on how badly it underperformed—because, unlike science, at a crucial turning point in its history it failed to accumulate knowledge. It seems hard to escape this conclusion, in this scenario, even if we do not know what account of scientific progress is correct. This is why I call this an unclear case for pessimism: this case does not support pessimism as strongly as a clear case for optimism supports optimism, but it does entail that philosophy is less successful than science, at least to some extent.

Of course, in this scenario, we would not be able to say whether philosophy made scientific progress in connection to the development of predicate calculus since it is debatable whether justification or knowledge is necessary for scientific progress. However, notice that having established that at a given juncture philosophy underperformed only with respect to justification and knowledge, no further historical question remains about that for the purposes of answering the exceptionalism question. That is, all that remains to do is to resolve the debate about scientific progress. Thus, although this result does not circumvent questions about scientific progress, it effectively compartmentalises historical questions about whether philosophy underperformed, with respect to the markers, and questions about what markers are identical to scientific progress.

These considerations indicate the advantages of this approach. Either this approach allows us to circumvent the problematic debate about the nature of scientific progress—assuming that one of the existing accounts is correct—or it allows us to give an evaluation of the degree to which philosophy has underperformed independently of the debate. Moreover, this approach allows us to compartmentalise factual questions about the success of philosophy and questions about the nature of scientific progress. This seems a clear advantage. Quite in general, it is good to be able to separate a problem into two research questions rather than having to do a combinatorial calculus to consider all possible combinations of answers to each question: epistemic pessimism, noetic moderate optimism, and so forth.

Another advantage of this approach is that it is more likely to lead to results that consolidate into a consensus. A result of the comparison between science and philosophy that considers each achievement associated with scientific progress can be accepted by participants in the debate irrespective of what view they take on the much-debated question about the analysis of scientific progress. By contrast, any view about whether philosophy made scientific progress is bound to be met by scepticism about the underlying definition of progress. That is, my proposal removes one possible point of attrition—the definition of scientific progress—on which philosophers tend to disagree by assessing whether philosophy underperformed with respect to several notable markers without committing to a view about which marker is identical with progress. If we are aiming to consolidate consensus, this seems desirable.

5 Conclusion

Dellsén et al., (2022) have argued that in order to sensibly discuss progress in philosophy, we need to first open, and ideally resolve, a debate about the analysis of philosophical progress. In this paper, I argued that although this is one possible approach, it is not the only viable and promising approach. The debate that Dellsén, Lawler, and Norton analyse focuses on whether philosophy makes the sort of progress that science makes: scientific progress. As such, we need not develop accounts of philosophical progress to give a framework to this debate; existing accounts of scientific progress can suffice. We could choose one account of scientific progress and then assess whether philosophy makes scientific progress by the standards of that account. But then our view would be only as good as the underlying account of scientific progress. Alternatively, I propose, we can address empirical questions about scientific progress in philosophy without choosing one definition of scientific progress. We can evaluate the success of philosophy by assessing if it has any of the markers associated with scientific progress, without choosing what definition of scientific progress is correct. Questions about the analysis of scientific and philosophical progress, as well as questions about the prevalence of philosophical but not scientific progress, remain open, but they have been effectively compartmentalised, allowing us to concentrate on one difficult question at a time.