In this book I have conducted an explicit and comprehensive case study of how one can apply reflective equilibrium (RE) as a method. Now it is time to take stock: what can we learn for RE and for how it can be used in philosophy? And what can we learn from applying RE for the subject of the case study, precautionary principles? In other words, this chapter addresses desideratum (6), which was identified in Chap. 1:

Desideratum 6:

The application and its results should be evaluated and critically discussed in order to learn from it for the use of RE as a method.

As we saw in the introduction, there are two fundamental worries about RE as a method. On the one hand, RE is seen as vacuous, or too permissive. According to Foley, RE is at best meta-advice: because it neither tells you what data are relevant, nor how to resolve trade-offs, it just “leaves you to muck about on these questions as best as you can” (Foley 1993, 128). On the other hand, there are also worries that RE is too demanding to be applicable. Because of its coherentist and holistic character, it might be unworkable for imperfect epistemic agents like us (cf. Van der Burg and Van Willigenburg 1998, 4):

[The] wide-ranging objectives of even a weak wide reflective equilibrium are at minimum intimidating and may be unattainable ideals of both comprehensiveness and coherence. (Beauchamp and Childress 2013, 405)

As a personal report, I can certainly confirm that applying RE can feel intimidating, and may sometimes involve some “mucking about”. However, the case study demonstrates that reflective equilibrium is a powerful methodological framework, which puts real constraints on the justification process and provides helpful guidance. In this chapter, I elaborate these points, work out further implications, and show how, based on the case study and its results, these two fundamental worries can be addressed. The discussion will stay on a rather general level, as results of the case study both for precautionary principles as well as for reflective equilibrium have already been summarized and discussed in detail at the end of each phase in Chaps. 6–8.

An important part of evaluating a method is asking whether it was suitable to pursue a given research goal—in this case, the justification of an action-guiding, moral precautionary principle. Thus, I start by discussing the main results for precautionary principles, and show how using RE contributed to them (Sect. 9.1).

I then focus on interesting results of the application of my specified RE method (in Sect. 9.2), before discussing more broadly what follows for RE as a method in philosophy (in Sect. 9.3). The case study shows that it is possible to specify RE as a method. However, I argue that it might be more fruitful to think of reflective equilibrium in the first place as a methodology, that is, a framework that guides decisions at various stages during the research process—including the selection of adequate methods. This also allows us to give a more satisfying answer to the two fundamental worries that RE is either too vacuous or too demanding.

9.1 Results of the Case Study for Precautionary Principles

The goal of this book was to explore what it would mean to seriously apply RE as a method, and to test whether this is both possible and fruitful. In the introduction, I distinguished between methods, methodology, and epistemology. I proposed to understand methods as concrete tools and techniques of research, in the sense of a set of instructions or steps which should be followed to achieve a given objective.

Consequently, whether the application of RE as a method was successful also depends on how well it was able to contribute to its objective. For the case study, the pragmatic-epistemic objective was to justify an action-guiding moral principle that is applicable to the subject matter of precaution and precautionary decision-making.

The results for precautionary principles are already discussed in detail at the end of each phase of the case study (Chaps. 6–8). This section therefore only provides a rough summary of how the Tripartite Precautionary Approach to Threats of Rights Violations (Rights-TPA) resulted from the application of RE, before outlining some possible next steps in the debate about PPs.

The Rights-TPA avoids the main objections PPs face, and is able to address the desiderata and open questions identified in Chap. 4. RE could significantly contribute to the formulation and justification of the Rights-TPA, but these are not the only benefits: the results of its application also allow us to gain further insights for the debate about PPs, and to identify possible next steps of research.

The Tripartite Precautionary Approach to Threats of Rights Violations (Rights-TPA):

  • The Rights Meta Precautionary Principle (Rights-MPP): Uncertainty must not be a reason for inaction when there are threats of rights implications.

  • The Precautionary Tripod: The elements that have to be specified in order to obtain an action-guiding precautionary principle version: If there is a threat that meets the harm condition (i.e., a specific rights violation) under a given knowledge condition then a recommended precaution should be taken.

  • Proportionality: Demands that the elements of the Precautionary Tripod are adjusted proportionally to each other, understood as Consistency: The recommended precaution must not be recommended against by the same PP version, and Efficiency: Among those precautionary measures that can be consistently recommended by a PP version, the least costly one should be chosen.

    The starting point for a rights-based PP version: If there is (1) a threat of a wrongful rights violation, then (2) select the least stringent knowledge condition that results in a consistently applicable version of PP given the harm condition. To comply with the Rights-MPP, uncertainty must neither render the PP version inapplicable nor lead to continual delay in taking measures to prevent rights violations.

The Rights-TPA is what I call an integrated PP interpretation, that is, it combines action-guiding, epistemic, and procedural elements (Steel 2015, 14). The procedural element, the Rights-MPP, puts constraints on the selection of adequate decision-rules: decisions must be made in a way that ensures that uncertainty about the likelihood or harmfulness of threats of rights implications will not lead to inaction. This means, for example, that we need effective ways to make decisions about protective measures also in situations in which no reliable probability information is available.

As Steel (2015, 18) argues, the MPP is a substantive and informative principle: for example, it will recommend against the use of decision rules that allow precautionary action only if it can be shown that the expected benefits of the precaution outweigh its expected costs. Such a rule would lead to inaction due to uncertainty when expected costs and benefits cannot be reliably forecast (Steel 2015, 21). Consequently, it will often speak against the use of standard approaches of risk management, like cost-benefit analysis.

The Rights-TPA thereby avoids two of the main objections PPs face: it is neither vacuous nor redundant. Due to the Proportionality-Element, it also avoids the other main objections, i.e., that PPs are incoherent and can lead to paralysis, and that they are irrational because they allow merely speculative harm to be a reason for strict regulations that might do more harm than the original threat (see Chap. 4 for the main objections against PPs). The Precautionary Tripod has to be specified on a case-by-case basis in a way that meets the constraints of both the MPP and the Proportionality-element. This means that measures have to be selected that do not themselves meet the knowledge and/or harm condition of the PP version, blocking both (a) the possibility that measures are taken that pose a greater threat than the one they are supposed to prevent, as well as (b) the possibility that a PP version leads to paralysis by justifying a measure while at the same time recommending against the same measure. Additionally, the Rights-TPA has the benefit that it is not tied to a specific category of uncertainty, but can be adjusted with respect to varying degrees of knowledge and available evidence. This ensures that all available evidence will be taken into account, instead of, e.g., solely comparing possible worst cases. In the RE process, this characteristic contributed to the superiority of the Rights-TPA to the Rights-Maximin-PP, leading to its adoption as the current system at step A6. For example, the Rights-TPA can account for more commitments than the Rights-Maximin-PP, and it does not require an additional criterion for “reasonable outcomes”. Which outcomes should be included, i.e., which threats should be treated as “realistic enough”, is built into the Proportionality criterion of the TPA: “reasonable outcomes” are those against which we still can take precautionary measures that do not themselves meet the harm condition and the knowledge condition of the threat.

The Rights-TPA is an adapted version of Steel’s (2015) proposal for a PP. The main difference from Steel’s proposal is the special attention paid to the protection of rights. This modification is one of the results of the RE process: at step A4 during the process of adjustments, it became apparent that none of the current candidate systems could account for a subset of the current commitments that could all be interpreted as distinctly moral value commitments. These concerned, e.g., the difference between risk-taking and risk-imposing, the protection of human health and the environment, the protection of rights of future generations, or paying attention to those that would be worst off if an uncertain harm should materialize.

One possibility would have been to exclude these commitments from the subject matter, e.g., by arguing that the target PP should only apply to decisions where the relevant values are determined in some other way. However, as the pragmatic-epistemic objective of the case study was explicitly to justify a moral principle, such a move would have meant abandoning the initial objective. Additionally, value-commitments such as that human health or the environment deserve special protection are arguably central to the subject matter of PPs (see Chap. 4). It is thus at least plausible to construct a PP that expresses substantial values instead of being a purely prudential or rational principle that tells us which means to take to achieve a given end. As argued in phase 3 of the case study, adopting the protection of rights as the normative basis for a precautionary principle provides a unifying rationale, since most if not all relevant cases of harm to the environment or to human health will be subsumable under it, as will be cases of threat of serious or catastrophic harm.

As I argue in Chap. 8, the fact that the Rights-TPA turned out to be the most convincing candidate also supports the original approach of Steel (2015), which is shown to be a very comprehensive and systematic formulation of a precautionary principle. Adapting Steel’s proposal to apply in particular to threats of rights violations makes it more substantial as a moral principle, which fits the pragmatic-epistemic objective of the case study. However, Steel can be interpreted to have the objective of formulating and defending a principle for public policy-making, and to refrain from making a commitment to substantial moral values or theories for this reason. Nonetheless, the Rights-TPA and the TPA are compatible to a certain extent: they will only lead to conflicting verdicts if the harm-condition of a PP-version of the TPA is set in a way that conflicts with the Rights-Meta-PP. Being clear about their relative objectives helps to see how they can both be precautionary principles, yet systematize different sets of resulting commitments.

This insight can be extended to the debate about precautionary principles more broadly. Talk about “the precautionary principle” without further qualification should be abandoned, e.g., statements about what “the PP” says, entails, or demands. Instead, it will be more fruitful to talk either about ideas commonly associated with PPs—that is, which can legitimately be seen as central commitments about the subject matter—or to refer to specific PP proposals that are defended with respect to a specific pragmatic-epistemic objective.

There might emerge one proposal that becomes authoritative at least for a specific context, making reference to “the PP” sensible in this context. However, at least given how diverse and at times fragmented the debate currently is, it is more fruitful to be as clear and explicit as possible.

Avenues for Further Development of PPs

This leads us to the question of what we can learn from the results of the case study for further research on PPs. As we have seen, the Tripartite Precautionary Approach to Threats of Rights Violations (Rights-TPA) is a candidate for a moral PP which can be defended based on the RE criteria: it answers common objections, can account for a broad range of commitments, and does justice to theoretical virtues.

However, this does not mean that the debate is settled now once and for all. The account of justification that RE provides is holistic, but we cannot consider and evaluate everything at once. One consequence of this is that we typically have to work with working hypotheses and assumptions that we need in order to “scaffold” other parts of the position, before we can go back and re-examine and elaborate them (which then, in turn, can of course have further implications for the rest of the position) (cf. Elgin 1996, 106; 2017, 20). Next steps with respect to the Rights-TPA would be to explicate the concept of “uncertainty”, and to elaborate the theory of rights that has to be assumed in the background in order to make the Rights-TPA applicable.

Another consequence is that it is always possible for further relevant considerations to emerge, which so far have been overlooked, and which unbalance the resulting position. In fact, this happened several times during the application of the RE method in the case study. This means that even a relatively stable position with a high degree of justification should never be seen as a final end point. Instead, I argue that each application of the resulting system is best conceived as an ongoing process of further elaborating, testing, and refining the position. The (Rights-)TPA has been shown to be justified to a high degree, but there is always the possibility that it leads to a verdict that we just are not willing to accept, i.e., which conflicts with a central commitment. Nonetheless, just because its justification is defeasible does not mean it is unreasonable to accept the Rights-TPA and use it to decide how we should proceed when facing threats of rights implications (see Chap. 2). At least as long as there is no other candidate that can be shown to be justified to a higher degree, we should commit to the Rights-TPA.Footnote 1

A useful way to work forward from here would be to apply the (Rights-)TPA systematically in applied ethics and case studies on policy recommendations. This would, on the one hand, mean that we can profit from the guidance of the (Rights-)TPA. On the other hand, this should be done in the spirit of RE, that is, we should see each such application also as an opportunity to further test and refine the system, and to search for potential problems and conflicts.

9.2 Results of the Case Study for the Method of Reflective Equilibrium

In this section, I focus on interesting results of the application of the RE method as it was specified in Chaps. 3 and 5. The next section will discuss more broadly what follows for RE as a method—or methodology—in philosophy. I start by summarizing how RE was specified as a method for the application in the case study, before highlighting interesting and relevant results from the application of this method.

9.2.1 Specifying an RE Method

In the first chapter, I introduced the distinction between epistemology, methodology, and specific methods. I roughly defined epistemology as a theory and analysis of what has epistemic value, methodology as a theory and analysis of how research should proceed, and methods as concrete techniques for conducting research (cf. Ackerly and True 2013).

Applying these distinctions to the present book, we can say that Chaps. 2 and 3 developed reflective equilibrium as a methodology which is based on an imperfect procedural epistemology that is weakly foundationalist and takes understanding as the goal of inquiry. Chapter 2 focused on the epistemological foundations of this methodology, discussing the conditions under which an epistemic position is justified. Chapter 3 focused on how this methodology can guide actual research: it highlights the challenges that need to be addressed, and the decisions that need to be made, if one wants to specify a method of reflective equilibrium that can be applied. The chapter provides guidelines for how one can proceed in order to obtain an applicable method, and identifies steps that will help to structure the research process. This is in line with the idea that a methodology guides decisions at various stages during the research process, e.g., which methods are chosen, and provides a defense of those decisions to one’s academic peers (cf. Ackerly and True 2013, 137).

Chapter 5 then demonstrated how a workable, specified method of reflective equilibrium can be obtained by addressing the challenges and decisions outlined by the so-described RE methodology. In particular, this required that we concretize the two steps of an RE process of alternatingly adjusting commitments and adjusting the system. This was done through specifying the RE criteria, e.g., spelling out “agreement between commitments and system” in the form of an account-function.

Importantly, describing the starting position of the RE application also included clarifying my pragmatic-epistemic objective for the RE application. As explained in Chaps. 2 and 3, how the RE criteria are specified and weighted also depends on the particular objective of the process of inquiry. In my case, I did set the pragmatic-epistemic objective as “Justifying an action-guiding moral principle that is applicable to the subject matter of precaution and precautionary decision-making”. This subject matter has been described in Chap. 4, which gave a survey about different interpretations of precautionary principles (PPs), along with justifications that have been brought forward for them as well as objections against them. Chapter 4 thereby also provided the basis for identifying relevant commitments, background elements, and candidates for the system.

Notably, the RE methodology as described in Chaps. 2 and 3 turned out to be elaborate enough to allow me to concretize RE as a method and to identify its input. This is not to say that this is the only way to specify RE as a method. In particular, I did work with some approximations and simplifications, where others might want to use more refined measures. For example, I only assigned rough ordinal weights of low–medium–high to commitments, but then de facto measured these weights on an interval scale when defining my account function in order to allow me to get an approximate comparison of how well different candidate systems can account for current commitments. There is no reason to think that this is the uniquely best, or even just one of the best, ways to measure these criteria. However, it shows that it is possible to operationalize the RE criteria in an applicable way, and thus lends support to the developed methodology. Further research and applications can now refine the ways in which RE can be concretized as a method, and draw on various epistemological theories to develop ways in which criteria like independent credibility of commitments or support from background theories can be spelled out more precisely.

9.2.2 Results from the Process of Adjustments

Once RE was specified as a method, the two steps of either adjusting commitments or adjusting the system could be applied to structure the process of adjustments. This application could, on the one hand, vindicate and illustrate certain aspects of the epistemological conception of RE. On the other hand, it also had some unexpected yet insightful results that help to further develop our understanding of RE.

Applying RE Can Fruitfully Guide the Development of Epistemic Positions

As in particular the first phase of the case study shows, RE is not only a method to achieve balance between considerations that are all already made explicit and developed. For example, I started by assessing two of the commitments—the Rio PP and the Wingspread PP—as candidates for principles, but they did not meet the RE criteria to a sufficient degree and were rejected. However, identifying their shortcomings allowed me to formulate guiding questions for the systematic exploration of commitments, and for developing improved candidate systems. Additionally, this illustrates that fleshing out the set of commitments, and filling in gaps, can also be an important result of an RE process—not only adjusting commitments under pressure from an existing, fully developed system, but also when searching for answers to open questions. This brings us to another insight, namely the important role of emerging commitments.

Emerging Commitments Are an Important Aspect of RE

Any application of RE as a method obviously depends partly on the input that we explicitly consider. As explained in Chap. 3, we typically do not have a complete overview of all our commitments, meaning that we have to work with a relevant selection,Footnote 2 and that we should continuously be on the lookout for further relevant commitments. Such emerging commitments, even though they were not initially explicitly considered, still count as input commitments, that is, they constrain the subject matter and have a degree of credibility that is independent of their agreement with the current system.

Throughout the process of applying the RE steps, emerging commitments turned out to play an important and sometimes decisive role. For example, when searching for further relevant commitments during step B5, several commitments emerged that caused problems for the Rights-Maximin-PP. In particular, as soon as the conditions of the Rights-Maximin-PP are met, it recommends that we select the course of action with the best worst case, disregarding additional information that we have, e.g., on the likelihood of the possible outcomes. However, based on arguments from the literature, I adopted the commitments EC 28 and EC 29 at step B5:

EC 28:

A decision principle for decisions under uncertainty needs criteria to decide which outcomes should still be included as “reasonable” or “plausible” enough. [medium] [emerged at Step B5]

EC 29:

The information we have about possible outcomes of courses of actions should not be irrelevant for the decision process only because it is not sufficient to assign reliable probabilities. [medium] [emerged at Step B5]

Both of these commitments are in conflict with the Rights-Maximin-PP, which led me to introduce the Tripartite Precautionary Approach to Threats of Rights Violations (Rights-TPA). The Rights-TPA can account for more commitments than the Rights-Maximin-PP, and it does not require an additional criterion for “reasonable outcomes”. Which outcomes should be included is built into the Proportionality criterion of the TPA: “reasonable outcomes” are those against which we still can take precautionary measures that do not themselves meet the harm and knowledge conditions of the threat.

A Dynamic and Non-Linear Process

That the Rights-TPA is chosen over the Rights-Maximin-PP at step A6 illustrates how in the course of a process, the acceptability of a candidate can change: at step A4, the TPA was rejected in favor of the Maximin-PP. At step A6, further information was made explicit, turning the scales in favor of the TPA.

This leads us to the more general point that the RE process is dynamic and non-linear. Firstly, previously rejected elements, like (parts of) a candidate system or specific commitments, can become acceptable later on, and vice versa. Secondly, re-adjustment of commitments is possible, meaning that commitments that were already adjusted can be re-adjusted (for example, at Step B6, C 13 replaced C 4 which had previously replaced the initial commitment IC 12). Thirdly, the set of explicitly considered commitments can change throughout the process. I have already stressed the importance of emerging input commitments. Another way in which the set of current commitments might be expanded is through inferences from the current system, like the newly inferred commitments in step B2. And fourthly, it is possible that parts of the background should move to the foreground and vice versa. For example, in phase 2, the explication of “Precautionary Measures” from phase 1, ExplicPrec, lost its relevance for the position in the foreground and could be moved to the background: the relevant candidate systems at this point, the RCPP and the Maximin-PP, will both only select measures that fulfill the criteria for being a precautionary measure, without needing ExplicPrec to yield a verdict.

“Sub-processes” are Possible, in which Only a Part of the Position is Adjusted

The explication of concepts, like “precautionary measures” in Chap. 6, or developing parts of the system, like the rights threshold in Chap. 8, are examples of “sub-processes” within the RE process. The case study shows that such sub-processes, in which only a part of the system and a subset of the commitments are adjusted with respect to each other, can be integrated into the RE process.

Relevance of the Background for the Position in the Foreground

What can or cannot reasonably be seen as part of the background has significant implications for the interpretation and adjustment of elements of the position in the foreground. Here are three especially important ways in which this can be the case: Firstly, whether or not a system can account for a commitment will often depend on the available background information, e.g., the factual information about a case. Secondly, whether or not a system has a theoretical virtue can also to some degree depend on background elements. See the example of the virtue of determinacy in Chap. 6: here, candidate systems have a low determinacy because terms that they use are not clearly defined in the background. Thirdly, how much weight should be given to a commitment will also partly depend on whether or not it can be supported with relevant background theories.

Functional Difference between Commitments and System

I argued that the relevant difference for RE is not between particular judgments and general principles, but between the attitude of commitment on the one side, and the ability to provide a systematic account on the other (see Chap. 2). There are different constraints on both sides, which is also specified in the steps for adjusting the system (A-Steps) and the steps for adjusting commitments (B-Steps) of the RE method. While a system has to be able to account for commitments while doing justice to theoretical virtues, commitments have to be in agreement with the system, have to respect input commitments, and must have some independent credibility. But none of this requires that commitments and a system have to be different from each other in content or in form.

The application of the RE method shows that this is a plausible way to draw the distinction. For example, in phase 1, the same propositions could be used both as commitments and as candidate systems. In other cases, whole classes, or subsets, of commitments were discussed and adjusted, e.g., whether all commitments referring to situations in which probabilities are available should be excluded from the subject matter. This shows that not only commitments to particular judgments are relevant, but also—and maybe even more so—general commitments.

Additionally, some ways in which a system fails to account for commitments can be especially insightful, e.g., it can become apparent that they all have something in common. For example, when adjusting commitments with respect to the Maximin-PP in step B4, one thing that became apparent was that the Maximin-PP could not account for a subset of the commitments that could all be interpreted as distinctly moral commitments. As the pragmatic-epistemic objective (see Chap. 5) was explicitly to formulate a moral principle, this is problematic for the Maximin-PP. However, the other candidates for the system up to this point faced the same problem. This issue was addressed in phase 3 through a sub-process in which the rights threshold principle was developed to supply the Maximin-PP as a part of the system.

Reasonable Constraints, but No Rigid Rules

Let me finish with a more general result on the kind of guidance that the RE method provides. While the method did not provide rigid rules that unequivocally determined which adjustments should be made, it proved to be applicable in an informative way that put real constraints on the process of justification. This is in line with how the epistemology and methodology of RE were described in Chaps. 2 and 3: trade-offs can be resolved in different ways, and there is no reason to think that there will always be a single best way to adjust a position. RE allows for reasonable pluralism, and a specification that does not include this would not be adequate.

That there was not always a single, unequivocally best way to adjust the position does not mean that the adjustments were arbitrary, either. Using the RE criteria to assess and compare candidate systems in the A-steps, or to assess commitments and different ways to adjust them in the B-steps, provided the basis for an informed choice, and to bring forward arguments in favor or against different ways to move forward. For example, comparatively assessing the Principle 3 System, the Rawlsian Core Precautionary Principle (RCPP), and the Utilitarian Uncertainty Principle (UUP) in phase 2 allowed me to clearly describe their respective strengths and weaknesses, and to make trade-offs explicit. On this basis, I could bring forward an explicit argument for adopting the RCPP.

Similarly, when adjusting the current commitments with respect to the RCPP at step B3, the RE criteria allowed me to identify which commitments are in conflict, and to assess how different ways to adjust them would influence the position. Commitments were not blindly adjusted in order to increase agreement with the system, but their independent credibility, expressed in form of the low–medium–high weights, had to be respected. For example, the following commitment:

EC 15:

Precautionary measures should not introduce serious threats of their own. [low] [emerged at Step B2]

Was replaced by the new commitment:

C 3:

Precautionary measures should not introduce threats that are equally or more severe than the threats they are aimed at, i.e., threats that have the same or a greater potential for harm. [medium] [replaced EC 15 at Step B3]

The reason for doing so was that this was a defensible way to increase agreement with the system. To avoid paralysis in cases where there is no completely safe option, one should not demand that precautionary measures never introduce any serious threat of their own. Arguably, the important point is that the overall threat level is reduced, making it reasonable to replace EC 13 through C 3.

However, there were no convincing arguments why cases where probabilities are available should be excluded from the subject matter, even though the current system at step B3—the RCPP—could not account for those commitments. Consequently, commitments referring to cases with available probabilities were not rejected, even though it would have increased agreement between the commitments and the system.

This exemplifies how applying the RE steps allows us to defend adjusting commitments and selecting candidate systems based on arguments. Using the RE criteria to assess and compare candidate systems, or to assess commitments and different ways to adjust them, provides the basis for an informed choice based on arguments in favor or against different ways to move forward. At the same time, these adjustments and the reasons behind them can be presented in a way that is traceable by others, which would allow them to go back in the process and to explore what outcomes one would reach if one were to make different choices.

9.3 Discussion: Reflective Equilibrium as a Methodology and Method in Philosophy

In the first chapter, we identified a range of challenges for applying RE as a method, most of them having to do with the fact that it was unclear what it would entail to apply it, e.g., what its input would be, how we should proceed when making adjustments, or how we could assess whether or not an actual state of RE was reached. Previous attempts at applying RE have not addressed these challenges to a satisfying degree. The case study in this book shows how they can be addressed and how an applicable method of RE can be obtained and put into practice. On this basis, we can now address the two fundamental worries that concern whether RE is either too vacuous, or too demanding, to be an applicable method.

The focus of this book was on applying reflective equilibrium as a method. In this last section, I want to take a step back in order to adopt a broader perspective. We saw that RE can be specified and applied as a method. But is this the most fruitful way to think about it and to make use of it? That is, if we draw the distinction between methodology and method in the way I suggested, does it make the most sense to see RE as a method in the sense of a set of instructions that should be followed in order to achieve a given objective? I argue that it might be more fruitful to develop and use RE as a methodology in philosophy—and that this will allow us to give a more convincing answer to the two worries.

I start by reconsidering some aspects of the case study that might cast doubt on whether RE is best understood as a method. The case study was specifically designed as a test for a step-by-step, open-ended application of the method of reflective equilibrium. In the form of the A- and B-steps and the stopping rule, the method provides guidelines that should be followed in a particular order. And Chaps. 6–8 describe a process of adjustments that develops a position through following these steps.

However, to what degree do these chapters describe what I actually did when constructing my position? That is, does the case study show that the RE method provides us with a set of explicit instructions that we can and should follow when constructing an epistemic position? Does it prescribe the exact steps we should take when we want to formulate and justify, e.g., a theory that does not yet exist?

In fact, the presentation of the process of adjustments in this book is to a significant degree a “cleaned up” reconstruction. For example, choices had to be made how to divide the process into three phases in order to be able to demonstrate and exemplify different aspects. Some of the confusing and ultimately misleading pathways have been left out in the final presentation of the process in order to achieve at least some comprehensibility for the reader. But even in this reconstruction, not everything strictly follows the logic of the A- and B-steps. Think about the sub-processes, like the explication in phase 1, or the formulation of guiding questions: while these could be integrated, and are interesting results, they also show that it was sometimes necessary to “tweak” the steps a bit to keep following the structure of the two alternating steps of going back and forth between commitments and a system. This “tweaking” does not directly speak against the possibility of spelling out an RE method—in fact, further developments of the method could try to incorporate this and allow for a more flexible structure. However, more generally, the dynamic and non-linear structure of the RE process casts doubt on the degree to which we can have a set of instructions that should be followed in a given order. Even though the case study shows that adjustments were not arbitrary, and RE provides helpful guidelines, this might not be enough to satisfy critics who see RE as too vacuous and permissive to count as a real method. They could argue that, after all, a lot depends on other philosophical methods that are applied as part of the RE applications, like deductive arguments, explications, or thought experiments (e.g., to make commitments explicit).

Once we take a step back from the case study, we see that it might be more fruitful to think of RE as a methodology, and not as a set of specific instructions that should be followed in a particular way. Let me elaborate.

Process Versus State of Reflective Equilibrium

If we understand RE primarily as a methodology, this raises questions about the role of the process of adjustments. If the process is not primarily understood in terms of describing a method by providing steps that should be followed, then what is its role? Why do we even need the process in the first place—would it not be enough to specify the conditions of having reached a state of reflective equilibrium? The important thing seems to be the resulting position and whether it can be defended. How we get there could be seen as a part of the process of discovery and irrelevant for justification. However, there are several reasons why we still should take the process seriously.

First, there are epistemological arguments for the process: it is correct that we should not conflate our psychological processes with our epistemic justification. The process of adjustments as part of RE is not intended as a description or prescription of actual cognitive practice (Baumberger and Brun 2021, 7936), but as a reconstruction. We do not have to recount in detail the genealogy of our position.

However, we need to be able to reconstruct a process of adjustments from the input to the resulting position in order to assess whether the resulting position is justified. As Baumberger and Brun (2021) argue, reconstructing such a process will often be the only way to decide whether a position is in reflective equilibrium. I see two main reasons for this: one is that the clarification and specification of the relevant configuration of epistemic goals is part of the process of equilibration (see also Chap. 3).

We may know, for example, that our pragmatic-epistemic objective demands that we do not give up too much precision in favor of simplicity, but how much we must finally give up is just as much a research question as the question of which commitments we will end up with in reflective equilibrium. (Baumberger and Brun 2021, 7937)

Another reason for not cutting the process out of the account of justification is that an RE state is not something that can be reached as a final product that is guaranteed to stay stable (cf. Bonevac 2004). We cannot completely survey all of our commitments, and it is always possible that new considerations should arise that unbalance the position, or that something changes in the background. Instead, we have a continuous progression of positions that are more or less justified, more or less stable, and connected through processes of adjustments. This does not render the RE state insignificant, however. It is still important as an ideal because it articulates what we are searching for. We can assess and analyze positions with respect to this ideal, thereby informing how the process can be continued.

That we can analyze positions with respect to this ideal does not mean that we know beforehand what the ideal position would be, otherwise we could simply adopt it. Instead, it means that the RE criteria can help us to identify potential weaknesses of our positions, i.e., where they are lacking with respect to an ideal RE state. In turn, this helps us to defend them, if, e.g., at the moment we cannot identify any problems.

Second, even if the process-aspect of RE is intended as a rational reconstruction and not as a description or prescription of how inquiry should proceed, it still is a powerful methodological framework. It will often be worthwhile to conduct one’s research along the lines of RE, e.g., to try to be explicit about what the input commitments are, to explicitly state the pragmatic-epistemic objective one pursues and what implications this might have for, e.g., the theoretical virtues one expects one’s system to have, and to document how one’s position changes. This provides useful heuristics and guidelines without forcing one’s thinking into a strict corset. At the same time, it will better enable one to later show via a reconstruction as an RE process that one’s resulting position can be defended as being in a state of reflective equilibrium. Discussing and defending results in this way will also help to identify gaps and avenues for further research, for example, which concepts still need to be explicated, identifying potential tensions between commitments, or between the position and background theories, etc.

Uses and Benefits of RE as a Methodology

Reflective equilibrium can provide a fruitful methodological framework for philosophical inquiry in different ways: it can be used as methodological approach to structure one’s pragmatic-epistemic projects, or to reconstructively appraise positions, but also to analyze debates and to compare positions. For example, one can analyze a debate in terms of different agents trying to bring more-or-less overlapping sets of commitments into different RE states, with potentially differing pragmatic-epistemic objectives. Such an analysis provides the basis for situating oneself in the debate, and to identify problems worth addressing.

Using the RE framework in this way would also help to connect seemingly isolated, small-scale projects to bigger epistemic projects and allows us to structure the academic division of labor. Each RE process, even if the RE conception is designed for a single epistemic agent, ultimately has to be seen as part of a collective effort.

One reason is that each RE process of developing a position in the foreground has to take place against a background of other theories, factual knowledge, and assumptions. Parts of this background, at least if they are supposed to support the position in the foreground, have to be defended based on their own RE processes, and those will often be conducted by other agents. Another reason is that it is desirable to develop many different positions, and to follow even those pathways that might not initially seem especially promising. They might turn out more successful later than initially anticipated. But, importantly, even if they turn out to be unfruitful, we have learned something more, and are better able to defend alternative positions that can be developed.

Understood as a methodology, RE does not prescribe particular methods or exact criteria no matter the question (cf. Ackerly and True 2013, 137). This fits with Walden’s (2013) defense of RE as non-essentialist, rejecting the idea that any kind of inputs or specific standards should be regarded as fixed. However, I argue that this characterization is much better suited for a methodology than for a method: a method that does not define any kind of specific standards could indeed rightly be accused of being vacuous and not helpful. A methodology, however, provides the framework to select appropriate methods for pursuing a given research goal, that is, for specifying criteria, for identifying and selecting input, for interpreting results, and so on.

Going beyond general appeals to RE and instead using it explicitly as a methodology promises to be highly beneficial for philosophical inquiry and debate. It not only provides useful guidance, but also forces us to make things explicit in a way that are specific advantages of RE. Let me name three examples. Firstly, it is important to acknowledge the functional difference between commitments and a system. They both have to meet different constraints, which we can only assess when we are clear about the respective roles. For example, assessments of theoretical virtues like simplicity only make sense for parts of a system, whereas asking whether something has independent credibility—e.g., through being intuitive—only makes sense if we are talking about commitments.

Secondly, differentiating between input and resulting commitments allows us to distinguish between a pre-systematic conception of a subject matter and a systematic account of it, which organizes and re-interprets parts of the subject matter in a new way. As the example of precautionary principles shows, this helps us to draw a distinction between commitments on precaution and precautionary principles that might be inconsistent, in tension, or simply unconnected to each other on the one hand, and developing a systematic account of this subject matter which might differ substantially from the initial conception.

Thirdly, the distinction between background and foreground forces us to make as explicit as possible what we presuppose. It also enables us to situate individual epistemic projects in a bigger context. By encouraging us to flesh out our positions, e.g., by systematically searching for further relevant commitments, applying RE will contribute to our understanding—even in situations where nothing is adjusted.

A Place for RE Methods?

I have argued that, ultimately, adopting reflective equilibrium as a methodology is more fruitful than to keep talking about it as a method. When adopting the RE methodology, some methods will be more appropriate for certain questions and subject matters than others: e.g., in political philosophy or applied ethics, working with questionnaires and taking into account folk commitments might be more appropriate than in, e.g., philosophy of science. A next step in research on RE could thus be to think about this more explicitly.

Also, I do not want to exclude that it is possible and useful in some sense to spell out an RE method. There might be certain areas, e.g., applied ethics or medical ethics, where it is possible to spell out an RE method to address particular problems. Especially in situations in which the RE process is relatively narrow and standardized, it seems likely that such a method could be specified (cf. Van der Burg and Van Willigenburg 1998, 13–15). What I here mean by “narrow” are cases in which there is, e.g., broad agreement on relevant background elements, and only a small part of the position in the foreground needs to be adjusted, which might be restricted to finding a solution to one specific problem—like selecting an appropriate treatment for a patient.

Summing Up and Answering the Two Fundamental Worries

I have argued that there is not a single, specific method of reflective equilibrium, consisting of specified standards on what kind of input to include, or rigid steps that should be followed. Instead, I argued that reflective equilibrium can be understood as a methodology which is based on an imperfect procedural epistemology that is weakly foundationalist. From this, it follows that objections to particular ways to spell out RE as a method—e.g., as taking intuitions as its input—do not necessarily amount to objections to the methodology more broadly. RE methods are specified with respect to a particular purpose, and in the spirit of the RE epistemology, are themselves revisable (Elgin 1996, 12). In a sense, reflective equilibrium is not a method among others; even specified RE methods might contain other things that are often counted as methods, like deductive inference, explications, or inductive arguments.

As an answer to the worry that RE is too vacuous to be a method, and at best can provide “meta-advice”, we can say that reflective equilibrium as a methodology is no more “meta” than other methodologies: it tells us how to approach processes of inquiry, and gives us guidelines for how to make decisions on what methods will be appropriate to address a specific problem of justification. In fact, it is an important element of RE that it does not impose rigid standards of justification, and takes seriously that we already have normative standards and theories that we have reason to regard as justified, and that should inform the ways in which RE is spelled out and applied.

As an answer to the worry that RE is too demanding, and will be overwhelming for imperfect epistemic agents, we can say the following: it is true that RE demands that we take many different things into account and that we balance them against various constraints. This certainly is demanding. However, just because something is demanding does not mean that it is implausible—justification is no trivial task. And, importantly, justification in reflective equilibrium is inherently provisional and an ongoing process, which takes the edge off this concern. Instead of being overwhelming and paralyzing, RE shows us how we can move forward and identify what can or cannot be justified at a given time.

As an answer to both worries, I have shown in my case study how the RE methodology can be successfully specified in a way that provides constraints which are informative without being arbitrary.