Philosophical Studies

, Volume 166, Issue 2, pp 243–256 | Cite as

In defense of reflective equilibrium

Article

Abstract

Recent years have seen a rekindling of interest in the method of reflective equilibrium. Most of this attention has been suspicious, however. Critics have alleged that the method is nothing more than a high-minded brand of navel-gazing, that it suffers from all the classic problems of inward-looking coherence theories, and that it overestimates the usefulness of self-scrutiny. In this paper I argue that these criticisms miss their mark because they labor under crucial misconceptions about the method of reflective equilibrium. In defending reflective equilibrium I put forward a handful of theses about the nature of inquiry (or, more generally, norm-governed enterprises) that form the backdrop to the method. The critics’ objections fall short, I argue, because they do not recognize reflective equilibrium’s embrace of these theses. Confronting these objections and understanding why they fail brings us to a better understanding what, exactly, the method of reflective equilibrium is. The answer I come to in the final section of the paper is that the method of reflective equilibrium is not, exactly, anything. It is a mistake to try to give a positive characterization of the view, to identify it with a concern with a particular species of data, particular procedures and methods, or even a particular conception of normative success. Instead, it should be understood as the denial of essentialism about just these matters—as a form of anti-essentialism about our epistemic inputs, methods, and goals.

Keywords

Reflective equilibrium Justification Coherence Normativity 

1 Introduction

Recent years have seen a rekindling of interest in the method of reflective equilibrium. Most of this attention has been suspicious, however. Critics have alleged that the method is nothing more than a high-minded brand of navel-gazing, that it suffers from all the classic problems of inward-looking coherence theories, and that it overestimates the usefulness of self-scrutiny.

In this paper I argue that these criticisms miss their mark because they misunderstand the method of reflective equilibrium. My defense occupies a hazy region between clarification and amendment. In defending reflective equilibrium I put forward a handful of theses about the nature of inquiry (or, more generally, norm-governed enterprises) that form the backdrop to the method. The critics’ objections fall short, I argue, because they do not recognize reflective equilibrium’s embrace of these theses.

Confronting these objections and understanding why they fail brings us to a better understanding what, exactly, the method of reflective equilibrium is. The answer I come to in the final section of the paper is that the method of reflective equilibrium is not, exactly, anything. It is a mistake to try to give a positive characterization of reflective equilibrium, to identify it with a concern with a particular species of data, particular procedures and methods, or even a particular conception of normative success. Instead, it should be understood as the denial of essentialism about just these matters—as a form of anti-essentialism about our epistemic inputs, methods, and goals. Practitioners of reflective equilibrium deny that we can say much of anything substantive in advance of inquiry, and they think of their “method” as whatever is left over of our ordinary thinking once we have purged these essentialist bogeys. In short, reflective equilibrium is what it isnt.

2 Goodman’s program

Before coming to the objections to the method of reflective equilibrium, it is worth quoting Goodman’s bravura statement of the doctrine.

The basic task in justifying an inductive inference is to show that it corresponds to the general rules of induction. […] But how is the validity of the rules to be determined? Here we encounter philosophers who insist that these rules follow from some self-evident axiom, and others who try to show that the rules are grounded in the very nature of the human mind. I think the answer lies much nearer the surface. Principles of deductive inference are justified by their conformity with accepted deductive practice. Their validity depends upon accordance with the particular deductive inferences we actually make and sanction. If a rule yields unacceptable inferences, we drop it as invalid. Justification of general rules thus derives from judgments rejecting or accepting particular deductive inferences. This looks flagrantly circular. I have said that deductive inferences are justified by their conformity to valid general rules, and that general rules are justified by their conformity to valid inferences. But this circle is a virtuous one. The point is that rules and particular inferences alike are justified by being brought into agreement with each other. A rule is amended if it yields an inference we are unwilling to accept; an inference is rejected if it violates a rule we are unwilling to amend. The process of justification is the delicate one of making mutual adjustments between rules and accepted inferences; and in the agreement achieved lies the only justification needed for either.1

These remarks express both of ideas typically associated with the method of reflective equilibrium. The first idea is a doctrine about the nature of epistemic normativity: a thing’s being justified, just, or valid is just a matter of its being the output of a suitably designed procedure. The other idea is a schematic account of this procedure. We reflect on our beliefs, desires, policies, principles, etc., scrutinize them, and try to bring them into an equilibrium that we can endorse from this reflective vantage point. Strictly speaking, these two ideas can be taken individually, but in practice they usually come as a pair.

In the next three sections I field three recent objections to the method of reflective equilibrium. Each of these objections fails, I suggest, because it falsely assumes the truth of a form of epistemic essentialism. The three kinds of essentialism are relevance essentialism (what sorts of considerations are relevant to our inquiries?), methodological essentialism (what methods ought we employ?), and criterial essentialism (what is the criterion of normative success?). My goal is to come to a better understanding of what the method of reflective equilibrium is (or, more importantly, what it isn’t) by responding to these objections.

3 The denial of relevance essentialism

Timothy Williamson says that we cannot even begin to evaluate the method of reflective equilibrium as a good or bad method without an account of the status and credentials of its inputs, of the things we start with and try to bring into equilibrium. But the defenders of reflective equilibrium punt on this question, Williamson says. They imagine the inputs of the method to be our “intuitions”, and think these things require no further justification or interrogation. And this means that they fail to offer a method we can even evaluate as good or bad.

A process generally acknowledged as at least superficially analogous to the attainment of reflective equilibrium in philosophy is the mutual adjustment of theory and observation in natural science. Imagine a description of it in which the word “observation” is used simply as a label for judgments with non-general content, irrespective of origin; it ignores the perceptual process. Such a description misses the point of the natural scientific enterprise. It provides no basis for an epistemological assessment. The nature of scientists’ evidence has been left unspecified. Similarly, one has no basis for an epistemological assessment of the method of reflective equilibrium in philosophy without more information about the epistemological status of the ‘intuitions’. In particular, it matters what kind of evidence ‘intuitions’ provide.2

As this passage makes clear, Williamson’s misgivings about the method of reflective equilibrium spring from a general antipathy for philosophers’ habit of resorting to reports about their ‘intuitions’. According to Williamson, this habit is a mistaken response to a problem about philosophical methodology. The legitimacy of disciplines like mathematics and the natural sciences is secured by their being grounded in evidence. Philosophy, it seems, should be no different. But in philosophical contexts much of our would-be evidence—things like “I have a hand”—is controversial. Indeed, our evidence is often a hair’s breadth away from the very thing we’re arguing about. This feature of philosophical discourse pushes philosophers to look for less vexed foundations, and, for many of them anyway, this means understanding our evidence in explicitly psychological terms. Instead of offering “I have a hand” as evidence, they turn to “I have a hand-like appearance before me.”

This is all a terrible mistake, Williamson says, because we can be just as mistaken about our own psychological states as we are about external states of the world. Thus we do not gain any revolutionary certitude by moving to more manifest starting points. But we do radically reduce the range of things we can prove on the basis of these starting points. So psychologizing our evidence is a bad deal. Williamson takes reflective equilibrium to be part and parcel of this turn toward psychologically manifest starting points. On his reading, reflective equilibrium construes the method of philosophy as trying to bring these special psychological appearances into harmony with each other; for him, reflective equilibrium is intuition mongering made methodical. But, as we just saw, the evidential status of our psychological appearances, our ‘intuitions’, can be just as dicey as ordinary statements about physical objects. So we cannot simply accept these things at face value and without further interrogation. The question we must ask, Williamson says, is what kind of evidence these intuitions are. Only then can we know whether or not the method of reflective equilibrium is a good one. This, however, is precisely the point where defenders of the method go mute.

Williamson gets the method of reflective equilibrium wrong in a subtle, but crucial way. By Williamson’s lights, reflective equilibrium involves a set fixed of inputs I and a method for equilibrating these inputs that will eventually yield answers to our questions. For most of his discussion Williamson deals with a version of reflective equilibrium whose inputs I are intuitions. But he acknowledges other writers, like David Lewis and Peter van Inwagen, who conceive of I as including mere opinions and beliefs.3 Now whatever the particular character of I, any defender of the method of reflective equilibrium worth his salt will insist that we include any considerations potentially relevant to the questions we are trying to answer, or at least those we have access to. Obviously if we leave out some relevant considerations, we are not going to get the right answer. But now Williamson comes along and asks about the “epistemological status” of I itself. Are the things in I evidence? Are we justified in taking them at face value? In asking this question, Williamson must assume that there is some further class of considerations, call them J, that are capable of answering it: the members of J are those considerations relevant to the higher-order question, “what is the epistemological status of the things in I?” But this is an odd assumption, for surely considerations that are relevant to this question will also be relevant to our original questions, to the questions that the members of I were trying to answer. If I am trying to figure out what time high tide is, and we include in I a particular tide chart, then a consideration in J to the effect that this “tide chart” is in fact a menu from Red Lobster is also relevant to our question of what time high tide is. It has the indirect relevance of an underminer. Call this property the transitivity of relevance: that if p is putatively relevant to a question, and q is relevant to the higher-order question whether p really is relevant, then q is also relevant to the original question. The transitivity of evidence is certainly an assumption of Williamson’s critique. His whole point is that if we find that the inputs to the method of reflective equilibrium lack epistemic credentials, then we are not justified in accepting the output of the method of reflective equilibrium.

Williamson’s critique therefore makes inconsistent assumptions about the method of reflective equilibrium. On one hand, charity demands that he grant that the inputs to the method of reflective equilibrium include everything potentially relevant to a particular question. (Failing to do that would mean neglecting the best versions of the view.) But on the other hand, his objection assumes that there are further considerations relevant to these questions—higher-order considerations—which reflective equilibrium neglects.

The assumption lurking beneath Williamson’s appraisal of the method of reflective equilibrium concerns how defenders of reflective equilibrium understand the “inputs” of that method. Williamson most commonly deals with a version of reflective equilibrium that takes intuitions as its inputs, and even when he acknowledges those writers, like Lewis and van Inwagen, who have a much broader conception of what reflective equilibrium takes in, he assumes that the method starts with some definite, fixed class of inputs. This assumption is crucial to Williamson’s case, for his criticism is that reflective equilibrium owes a story about the “epistemological status” of a particular class of inputs—like our intuitions about a given case. So Williamson assumes that the method reflective equilibrium embraces a thesis that I will call relevance essentialism.

Relevance essentialism: We can characterize what considerations will be relevant to an inquiry independently of that inquiry.

But this assumption is false. If we look back at Goodman, we see that he offers no positive characterization of what sort of states we are trying to bring into equilibrium. He does not say we ought to bring our intuitions, our instincts, or even our opinions into alignment. Supposing that Goodman does deny relevance essentialism, what does this denial come to? Denying relevance essentialism means refusing to bar anything from the method of reflective equilibrium a priori, and so affirming that any consideration—psychological episodes, beliefs, hunches—could potentially be relevant to a question we are entertaining. More concretely, it is to say that what counts as evidence in an inquiry is not settled a priori, but in the midst of that inquiry. It is in this spirit that Goodman talks about principles and inferences we are ‘unwilling’ to revise. He does not mean to suggest that our unwillingness to do without something endows it with some foundational status in our inquiries—that it gets to be an unquestioned datum. Rather, he is suggesting that reflective equilibrium gives us advice in mediis rebus—advice conditioned on our present engagement in inquiry and our having provisional ideas about which principles we are willing to revise, which inferences we won’t give up, and what things count in favor of what other things. Goodman formulates his advice in this way because he thinks that it is the only way to give epistemic advice. The idea of offering a theory of evidence per se amounts to doing epistemology from some de novo starting point, and that is impossible.

Perhaps it is unfair to say that Williamson is misunderstanding reflective equilibrium. Maybe he is assuming that relevance essentialism is true and trying to offer the most charitable reading of reflective equilibrium given that assumption. Is this assumption warranted? Williamson offers some remarks that could be taken as a defense of the claim, though I don’t know whether he intended them as such. He says that what makes physics, mathematics, and philosophy disciplines is that they are constrained by evidence.4 This claim strikes me as true, but anodyne. More importantly it is ambiguous: it could mean that there exists some definite thing, the evidence, that constrains these modes of inquiry from the very beginning, or it could mean that inquiry is constrained by various factors that we recognize in the midst of inquiry and come to retrospectively call “our evidence”. The proleptic reading of the claim is perfectly compatible with the rejection of relevance essentialism. (And it also strikes me a compatible with Williamson’s view; if it is, then the reflective equilibrium theorist and Williamson’s dispute is that much smaller than first appearances suggested.)

So what of the first reading? There are good reasons to reject this thesis. One is the kind of argument from the transitivity of higher-order relevance we saw above: if we keep supplementing the set of things relevant to an inquiry with those things that are relevant to the question of what is relevant to the question of what is relevant to the question (and so on), then it is likely we will end up with no restriction on relevance whatsoever. Another is the common thought that evidence is theory-laden, so we cannot hope to specify in advance of theory what our evidence will consist in, at least not with any substance. Finally, there are numerous occasions on which it is natural to say that not only has our evidence changed, but the sort of things we count as evidence have changed. At one time the rate at which other planets in the solar system move through sky would have been thought irrelevant to questions of terrestrial ballistics, but when Newton discovered that the same force governed terrestrial and celestial bodies, it became relevant.5

It appears that Williamson’s critique of the method of reflective equilibrium is either inconsistent or uncharitable. His basic mistake is to suppose that the defenders of the method of reflective equilibrium accept relevance essentialism, when in fact they have principled reasons for denying it.

4 The denial of methodological essentialism

Thomas Kelly and Sarah McGrath have recently given new voice to an old argument that reflective equilibrium offers a bad or, at least, inadequate method. Kelly and McGrath argue that one could be in reflective equilibrium but still be epistemically compromised.

On a traditional conception of justification, a belief is justified just in case it would amount to knowledge provided that it is true. Thus, to say that we are justified in believing that not everyone alive today will still be alive in 50 years time, is to say that our basis for thinking that this proposition is true is sufficiently strong that our belief qualifies as knowledge provided that it is true. Offhand, however, Goodman-justification looks too weak to underwrite genuine knowledge. After all, in principle, there is nothing that precludes the possibility that an inductive principle that passes all of Goodman’s tests with flying colors is in fact highly unreliable.6

And a few pages later, after claiming that the method of reflective equilibrium is a “dynamic coherence theory”, they say that

An individual might maintain a perfectly coherent set of beliefs while being completely unresponsive to relevant and easily perceptible changes in his or her environment. This is the point exploited by stock counterexamples to coherentism about justification in the epistemological literature. Intuitively, an individual who simply maintained the same perfectly coherent set of beliefs about her environment, despite the fact that her experiences of that environment were constantly changing, would not be justified in holding those beliefs (p. 333).

I am not altogether certain which ‘stock counterexamples’ to coherentism Kelly and McGrath have in mind. They may mean coherent grue-projectors or counterinductivists, but from the discussion they offer, it seems more likely that they have something more pedestrian in mind, something like the following case. Tabby is a scrupulous scientist, and maintains a strict regimen of equipment maintenance. Her methods, including the principle telling her that she can safely rely on her microscope, are kept in meticulous equilibrium. One day Tabby observes that her microscope is broken, but, for whatever reason, this observation does not issue in any new beliefs, so Tabby makes no effort to revise her results or conduct her experiments again. In this case Tabby maintains a “coherent set of beliefs about her environment, despite the fact that her experiences of that environment [are] constantly changing.” Kelly and McGrath assume that the defenders of reflective equilibrium would deem Tabby justified in her beliefs, but I don’t see why. Tabby is a scrupulous scientist, and so she must have a standing directive to integrate observations about her equipment into her book-keeping and into her beliefs more generally. If Tabby observes her microscope’s malfunction and does nothing, she is running afoul of this directive, and so she is out of equilibrium. To put the point another way, insofar as she has this directive, reflective equilibrium will demand that she made compensatory adjustments in her web of belief.7

Reflective equilibrium has a perfectly banal diagnosis of this case. Why do Kelly and McGrath think otherwise? Their assimilation of reflective equilibrium to doxastic coherence theories offers some clues. “The method of reflective equilibrium,” they say, “when understood as a dynamic coherence theory, does not seem particularly plausible as an account of how empirical scientists should arrive at their views of how the world works, given that it makes no essential reference to observation or perception” (p. 333). Kelly and McGrath seem to assume that the equilibrium in question is one of beliefs and beliefs alone, so my suggestion that Tabby is thrown out of equilibrium by something other than a belief—an observation, a test result, a standing directive—is not possible. But there is no reason to assume that reflective equilibrium theories want to circumscribe the range of equilibrium so narrowly. In point of fact, reflective equilibrium is not a coherence theory of the sort Kelly and McGrath describe. First, while some coherentists may believe that justification is simply a matter of harmony amongst beliefs, no advocate of reflective equilibrium will propose such a circumscription on the items to be brought into equilibrium.8 As we saw with the denial of relevance essentialism, they want the method to be as broad-minded as necessary about what things are to be brought into equilibrium as well as about what ‘equilibrium’ amounts to. Second, theories of reflective equilibrium are not wholly inward-looking in the way Kelly and McGrath intimate. Any decent account of reflective equilibrium involves some interface between our inner mental states and the outside world. This connection offers a constant source of new information that upsets partial equilibria and forces us to revise our understanding of the world. For example, on Catherine Elgin’s theory, the most sophisticated account of epistemic justification in the reflective equilibrium tradition, our picture of the world is being constantly bombarded by what she calls ‘initially tenable judgments’, many of which are perceptual in origin.9 Third, the method of reflective equilibrium is a social epistemology. Reflective equilibrium relies on Joey telling Tabby that her microscope is malfunctioning because it conceives of them engaged in inquiry together trying to bring their common judgments into equilibrium. It does not think of us as locked in our own private struggles for inner coherence, and it does not identify justification with such coherence.

Kelly and McGrath say that reflective equilibrium “does not seem particularly plausible as an account of how empirical scientists should arrive at their views of how the world works, given that it makes no essential reference to observation or perception,” and for this reason it is vulnerable to ‘No Contact With Reality’ objections. The word “essential” is doing more work in this claim than readers may appreciate. Any account of reflective equilibrium for members of the kingdom Animalia will most assuredly have plenty to say about perception. Elgin’s case is a good example of that. But that is because perception happens to be a very important source of information for animals. What proponents of the method of reflective equilibrium deny is that the deliverances of perception have a special, or essential, status independent of or prior to the scrutiny of reflection. They think perception must—and does!—earn its stripes in the same way as our other sources of knowledge. The disagreement between Kelly and McGrath and reflective equilibrium theories, then, is not about whether perception is important. Everybody agrees to that. The disagreement is over whether the role of perception in the formation of beliefs is something that we can work out in advance of our inquiries, whether, that is, our methods make “essential” reference to perception, or just turn out to rely on perception because we learn that perception is, in fact, a good source of information.

Reflective equilibrium theories deny essentialism about perception because they deny all forms of methodological essentialism.

Methodological essentialism: We can determine independently of inquiry which methods we must use in that inquiry, e.g. that we must rely on perceptual judgments.

Here there is a close analogy with the denial of relevance essentialism. Just as we cannot be sure in advance of inquiry what things will be relevant to that inquiry, we cannot be sure in advance what our methods will look like. Perhaps we should rely on perception in forming our beliefs, perhaps not: we have to see whether perception is a reliable method.10

With reflective equilibrium’s position properly understood, Kelly and McGrath’s criticism rings hollow. Saying that reflective equilibrium “does not seem particularly plausible as an account of how empirical scientists should arrive at their views” because it does not make “essential reference” to perception is rather like saying that some theory of scientific reasoning cannot be a particularly plausible account of physics because it does not make “essential reference” to large hadron colliders. The same reply is in order for both criticisms: we do not make essential reference to perception or particle colliders because we are not sure, in advance, whether they are good methods or bad. But when and if these facts are ascertained, they will certainly affect our methods.

5 Criterial anti-essentialism

Christine Korsgaard does not call her theory of normativity a version of the method of reflective equilibrium. But I class her alongside Goodman and company for a few reasons. She uses the nearby concept of ‘reflective endorsement’. Her debts to Rawls, and therefore to Goodman are obvious. And, most importantly, her theory is a natural generalization of Goodman’s. Whereas Goodman thinks that reflective equilibrium is the mark of epistemic justification, Korsgaard thinks that reflective endorsement is constitutive of normative success writ large.11 Korsgaard’s theory is also subject to similar criticisms. These remarks from Hilary Kornblith are representative:

In his first inaugural address, Bill Clinton offered an extremely optimistic view of the United States: “There is nothing wrong with America that can’t be cured by what is right with America.” A similar view about human reason seems to underlie Korsgaard’s account of reflective endorsement. Human beings are susceptible to certain sorts of bad reasoning. At the same time, we are able to discover that some of the reasoning to which we are prone is bad, and we do this, at least in part, by reflecting on the character of our reasoning. So what is good in our reasoning seems to work to correct what is bad. If we were to continue this process, “all the way back,” would we end up with proper reasoning? That is certainly one possibility. But it is an empirical bet, and a highly controversial one at that. […] Even on the extremely optimistic view that human reason would fully correct itself, it would be a mistake simply to identify good reasoning a priori with whatever might be the outcome of the self-reflective process.12

These criticisms are problematic. Kornblith seems to assume that there is some fixed standard of justified reasoning external to an agent’s own deliberations, one that those deliberations are aiming to achieve. And he assumes that Korsgaard is putting forward the method of reflective endorsement as a means to securing this state. But Korsgaard does not think that reflective endorsement is instrumentally useful in securing some further standard of justification; she thinks reflective endorsement is constitutive of normative success. The problem of the normative, she says, springs from “our capacity to turn our attention onto our own mental activities, [which] is also a capacity to distance ourselves from them, and to call them into question.” This conception of the normative problem reflects a Goodmanian pragmatism about normativity. Normative questions have a distinctive origin: our ability to reflect on ourselves and our actions. Korsgaard’s suggestion that justification is a matter of reflective endorsement emerges directly from this understanding of the normative problem. “If the problem springs from reflection, then the solution must do so as well. If the problem is that our perceptions and desires might not withstand reflective scrutiny, then the solution is that they might.”13
In other words, Korsgaard rejects instrumentalism. For her, the method of reflective equilibrium is not a method for securing some further normative goal that we can articulate in advance of normative thought. It is constitutive of normative success. The rejection of instrumentalism follows from the rejection of a broader doctrine I will call criterial essentialism:

Criterial essentialism: We can determine what is good in the way belief (or other actions and states) in advance of inquiry (or other activities), and our methods ought to be designed to secure these things.

Why do Korsgaard and other defenders of reflective equilibrium deny criterial essentialism? Their reasons emerge most clearly in the critique of rival theories. As it happens, Kornblith himself advances a theory of epistemic normativity. I will take it as a convenient example. On Kornblith’s view, we should adopt whatever methods are conducive to true beliefs because true beliefs make us better able to satisfy our desires (whatever our desires might be).14 Such a theory of normativity raises two questions: (a) Why suppose that epistemic reasoning does not have its own standard of correctness, independent of what it may do to help us satisfy our desires?, and (b) Why does the normative buck stop at our desires? Can’t we wonder whether we ought to have a certain desire and, more pressingly, whether it really gives us a reason to act? As far as I can tell, Kornblith doesn’t answer these questions.
For Korsgaard, these doubts about the normative authority of our desires are not a malignant form of skepticism, but the essence of normative thought.

I perceive, and I find myself with a powerful impulse to believe. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I believe? Is this perception really a reason to believe? […] The reflective mind cannot settle for perception and desire, not just as such. It needs a reason. Otherwise, at least as long as it reflects it cannot commit itself or go forward (p. 93).

This distinctive ability for detachment and scrutiny is the origin of normative thought, the place where the difference between reasons and desires emerges. Thus Korsgaard’s reflective endorsement view of normative in particular and the rejection of criterial essentialism more generally arise quite naturally out of the challenges that a critical agent will naturally pose to a view of normativity like Kornblith’s: the unrestricted reflective scrutiny that characterizes normative thought (that I can question why my desires give me reasons, for example) is incompatible with a view, like criterial essentialism, which sees certain normative facts as beyond question.

Kornblith’s objection to Korsgaard misses the mark because it neglects this point. Kornblith seems to interpret Korsgaard as offering a method that is instrumentally useful for achieving a given end, when in fact she thinks that normativity arises out of our ability to question any such end. This feature of Korsgaard’s view brings us back once more to Goodman. Goodman said that the justification of our inductive inferences is a pragmatic endeavor that survives our realization that we cannot offer an absolute guarantee that our predictions will turn out true. Korsgaard thinks the same about normative questions in general. Both Korsgaard and Goodman are therefore led to reflective equilibrium by the thought that it is the only conception of normativity that we are left with once we discover that a kind of absolutism is impracticable. It is the answer we are left with when we find that we cannot divine the future or use some inner sense to tell right from wrong.

Furthermore, the reflective equilibrium theorist is all for the integration of the scientific results that Kornblith cites in his case that Korsgaard’s ‘bet’ is a dubious one. Her proviso is simply that these results must be duly integrated into the normal course of inquiry—who could deny that the human tendency to rationalization and cognitive dissonance is irrelevant to our reasoning?—but that they not stand outside of it as some privileged, insulated, and final arbiter. This is the difference between a methodological naturalism on which our philosophical methods are continuous with our scientific ones, and a Procrustean naturalism which thinks our epistemology must be fit inside the rigid framework the natural sciences give us.

6 Reflective equilibrium as anti-essentialism

I have surveyed a handful of recent critiques of the method of reflective equilibrium. My rejoinders to these critiques have had more or less the same structure. Each of the critiques assumes a form of essentialism about our inquiries (or, in the case of Korsgaard, our activities more generally). By essentialism I mean the idea that we can determine the nature of certain facets of these inquiries in advance of the inquiries themselves, and that nothing that comes about in inquiry will change those facets. Relevance essentialism is the view that we can determine what kind of considerations will be relevant to our inquiries—what our evidence will be—in this way. Methodological essentialism is the view that we can determine the general shape of our methods in advance, for instance we can know that perception must play an essential role in those methods. Criterial essentialism is the view that we can determine a criterion of normative success in advance, for example that normative success involves satisfying desires.

These critiques fall short because the defenders of the method of reflective equilibrium reject all these essentialisms. Williamson’s charge is that reflective equilibrium theories are wrong about what is essentially relevant; Kelly and McGrath’s is that they are wrong about what is essential to our methods; and Kornblith’s is that reflective equilibrium does a poor job of bringing about the essential standard of normative success. But reflective equilibrium theories cannot be guilty of these errors because they deny all these forms of essentialism, a feature of the view we see this as far back as Goodman and his carefully non-committal characterization.

Surveying these critiques has led us to a better understanding of what the method of reflective equilibrium is, or rather, what it is not. The method of reflective equilibrium should be understood not as a particular method that lays out theses about the nature of evidence, methodology, and criteria for epistemic success, but as what is left over of our methods when we deny that we can do this.15

Such a characterization might lead some to say that if I am right, then the method of reflective equilibrium is a vacuous conception of our methodology, and so there is no point in articulating such a conception. But this would be a mistake. As long as there is a philosophical temptation to take some common, well-confirmed, but nonetheless contingent feature of inquiry and cast it as the very essence of inquiry, reflective equilibrium, understood as the negative thesis I have articulated here, will offer a tonic corrective.16

Footnotes

  1. 1.

    Goodman (1955, pp. 61–62).

  2. 2.

    Williamson (2007, p. 244).

  3. 3.

    Lewis (1983, p. x), van Inwagen (1997, p. 309).

  4. 4.

    Williamson (2007, p. 209): “In most intellectual disciplines, assertions are supposed to be backed by evidence. Mathematicians have proofs, biochemists have experiments, historians have documents. You cannot just say whatever you happen to believe. […] Let us proceed on the working hypothesis that evidence plays a role in philosophy not radically different from its role in all other intellectual disciplines. Without such a role, what would entitle philosophy to be regarded as a discipline at all?” Williamson’s question is rhetorical, of course, but I think we get a good answer from Kant. In the Metaphysical Foundations of Natural Science, Kant defines “a science” as “a doctrine that is supposed to be a system, that is, a whole of cognition ordered according to principles” (4:468). For Kant, something qualifies as a science on the basis of internal structural features—systematicity—and not because its inputs have the special status of evidence. Kant’s reasons for this are rooted in his metaphysics: securing the deep vindication that separates Williamson’s conception from Kant’s would mean knowing about things in themselves.

  5. 5.

    Williamson does have replies available to these points, of course. We could say, for example, that we were mistaken about what our evidence was before Newton. I worry that this factive sense of ‘evidence’ doesn’t fit with the way that we use the word, and it threatens to deny the concept its characteristic role in guiding scientific debates. This debate cannot be resolved here, obviously.

  6. 6.

    Kelly and McGrath (2011, p. 331). The objection I survey here is actually preceded by another. Kelly and McGrath quote Goodman as saying that we cannot have inductive knowledge. But, they insist, we clearly do have such knowledge. So there must be a notion of justification that goes along with inductive knowledge, a notion that solves the equation ‘x + true belief = knowledge’. Because Goodman-justification falls short of this standard, it also falls short as a notion of justification. But this is not what Goodman is denying when he says we have no knowledge of the future. His point, inelegantly stated perhaps, is that there is no way to gain absolute certitude about the future—“prevision” he calls it—and so our inductive methods are irredeemably fallible. The mistake he impugns in others is the assumption that justifying an inductive inference requires overcoming this fallibility with something that guarantees, come hell or high-water, that an induction is truth-preserving. Even if there is nothing we can do in our theorizing to guarantee the truth of an inductive conclusion, Goodman says, there is still a reasonable, and very pressing, question about which inductive inferences we ought to accept. Thus for Goodman the method of reflective equilibrium is attached to a fallibilistic conception of justification, but there is no reason to think it must be any weaker than that.

  7. 7.

    Now, if Kelly and McGrath respond to this suggestion by revising the case so that Tabby has no such standing directive to take on board observations, then the example becomes less comprehensible, and we are owed a more elaborate account of Tabby’s cognitive history to understand the case. For instance, we would need to know what forces led Tabby to adopt such a strange stance, and just how eccentric Tabby would have to be to wind up in such a position. As Sharon Street (2009) has recently argued, many of philosophers’ favorite coherent-but-strange characters trade on strategic under description, and these characters become less strange when we fill out their background stories in ways that explain their eccentricities. It is also possible that Kelly and McGrath’s stock counterexample is supposed to be one where an agent is completely isolated from indications of her unreliability: it is not merely that she is missing a belief that knocks her out of equilibrium; she doesn’t have an observation, an experience, a test result, a computer simulation, or any other acquaintance with this unreliability. So we stipulate that Tabby’s methods are unreliable but this unreliability is inaccessible to her. If this is the case, then the objection afflicts not just reflective equilibrium theories, but all forms of internalism, and the standard rejoinder seems to apply: insofar as terms like ‘justified’ are normative, they must be capable, in principle, of guiding our actions, so any example that forfends such guidance by fiat is moot.

  8. 8.

    This stance is probably rarer than is usually appreciated. Even Davidson (2001, p. xvi), in explaining how widely misunderstood his view is, declares that, “ ‘experience’ and ‘perception’ are perfectly good words for whatever goes on in our minds when we look around us, smell, touch, hear, and taste. I was so eager to get across the idea (for which I should have given credit to Wilfrid Sellars) that epistemic intermediaries between the world and our beliefs are a mistake that I made it sound to many readers that I were repudiating all serious commerce between world and mind. In truth my thesis then as now is that the connection is causal and, in the case of perception, direct.”

  9. 9.

    See Elgin (1996, p. 110ff).

  10. 10.

    One could argue that some methodology—probably one involving perception—must have an a priori warrant, or we will enter into a justificatory regress that ends in skepticism. But defenders of reflective equilibrium will deny this, and this doesn’t seem to be Kelly and McGrath’s argument anyway.

  11. 11.

    Korsgaard (1996).

  12. 12.

    Kornblith (2010, pp. 17–18).

  13. 13.

    Korsgaard (1996, p. 93).

  14. 14.

    Kornblith (2004, Chap. 5).

  15. 15.

    Sometimes the method of reflective equilibrium will have a more definite character. John Rawls’s (1971) political use of the notion is an example. Rawls specifies a particular kind of judgment, “considered convictions”, as the input of his procedure because that kind of judgment is the locus of the overlapping consensus that he is after. His reliance on considered convictions is an acknowledgment of a pragmatic constraint on the use of reflective equilibrium in political philosophy, namely that a particular kind of consent is required for a theory of justice to have authority. The particulars of Rawls’s use of reflective equilibrium therefore arise from the particulars of his project, not from abstract considerations about the essence of reflective equilibrium.

  16. 16.

    For helpful comments and discussion, I thank Catherine Elgin, Sophie Horowitz, Paulina Sliwa, and Ekaterina Vavova.

References

  1. Davidson, D. (2001). Subjective, intersubjective, objective. New York: Oxford University Press.CrossRefGoogle Scholar
  2. Elgin, C. Z. (1996). Considered judgment. Princeton, NJ: Princeton University Press.Google Scholar
  3. Goodman, N. (1955). Fact, fiction, and forecast. Cambridge, MA: Harvard University Press.Google Scholar
  4. Kelly, T., & McGrath, S. (2011). Is reflective equilibrium enough? Philosophical Perspectives, 24, 325–359.CrossRefGoogle Scholar
  5. Kornblith, H. (2004). Knowledge and its place in nature. New York: Oxford University Press.Google Scholar
  6. Kornblith, H. (2010). What reflective endorsement cannot do. Philosophy and Phenomenological Research, 80(1), 1–24.CrossRefGoogle Scholar
  7. Korsgaard, C. M. (1996). The sources of normativity. New York: Cambridge University Press.CrossRefGoogle Scholar
  8. Lewis, D. K. (1983). Philosophical papers (Vol. 1). New York: Oxford University Press.Google Scholar
  9. Rawls, J. (1971). A theory of justice. Cambridge, MA: Harvard University Press.Google Scholar
  10. Street, S. (2009). In defense of future Tuesday indifference: Ideally coherent eccentrics and the contingency of what matters. Philosophical Issues, 19, 273–298.CrossRefGoogle Scholar
  11. van Inwagen, P. (1997). Materialism and the psychological-continuity account of personal identity, in J. Tomberlin (ed.). Philosophical Perspectives, 11, 305–319.Google Scholar
  12. Williamson, T. (2007). The philosophy of philosophy. Malden, MA: Blackwell.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V. 2012

Authors and Affiliations

  1. 1.Department of PhilosophyDartmouth CollegeHanoverUSA

Personalised recommendations