INTUITION AND THE JUNCTURES OF JUDGMENT IN DECISION PROCEDURES FOR CLINICAL ETHICS

Theoretical Medicine and Bioethics (2007) 28:1 30 Ó Springer 2007 DOI 10.1007/s11017-007-9027-z INTUITION AND THE JUNCTURES OF JUDGMENT IN DECISION PROCEDURES FOR CLINICAL ETHICS ABSTRACT. Moral decision procedures such as principlism or casuistry require intuition at certain junctures, as when a principle seems indeterminate, or principles conflict, or we wonder which paradigm case is most relevantly similar to the instant case. However, intuitions are widely thought to lack epistemic justification, and many ethicists urge that such decision procedures dispense with intuition in favor of forms of reasoning that provide discursive justification. I argue that discursive justification does not eliminate or minimize the need for intuition, or constrain our intuitions. However, this is not a problem, for intuitions can be justified in easy or obvious cases, and decision procedures should be understood as heuristic devices for reaching judgments about harder cases that approximate the justified intuitions we would have about cases under ideal conditions, where hard cases become easy. Similarly, the forms of reasoning which provide discursive justification help decision procedures perform this heuristic function not by avoiding intuition, but by making such heuristics more accurate. Nonetheless, it is possible to demand too much justification; many clinical ethicists lack the time and philosophical training to reach the more elaborate levels of discursive justification. We should keep moral decision procedures simple and user-friendly so that they will provide what justification can be achieved under clinical conditions, rather than trying to maximize our epistemic justification out of an overstated concern about intuition. KEY WORDS: casuistry, decision procedure, discursive justification, heuristic, intuition, judgment, moral epistemology, moral theory, principles, principlism INTRODUCTION How should clinical ethicists decide particular moral cases? We could dispense with principles or paradigm cases and simply judge a case using intuition alone. In easy or obvious cases that works well enough, but in less obvious cases we sometimes we turn to moral decision procedures for guidance. Many of these procedures involve applying something to the case; principlism, the common morality method, the four-topics method, and specificationism all apply principles or other norms, while casuistry compares the case with

2 various paradigm cases. Procedures that apply principles or compare paradigm cases require intuition at what I will call the junctures of judgment : when it is not clear whether a principle or paradigm case applies to the case being decided, when the principle is indeterminate and it is not clear how it applies, when more than one principle seems to apply but they mandate conflicting results, or when we judge which paradigm case is most relevantly similar to the instant case. Intuition is frequently decisive at the junctures of judgment; that is, our conclusion about the case may vary dramatically with our intuitions about how the principles or paradigms relate to it. This worries many ethicists, for reasonable moral judges often have different intuitions, reaching different conclusions about the same case using the same decision procedure, but cannot justify their respective intuitions beyond simply asserting to one another that it seems so to me. This gives rise to an apparent dilemma: If moral intuitions lack sufficient epistemic justification, then moral decision procedures that use principles or paradigm cases are to that extent unsound. If intuitions are sufficiently justified, then principles and paradigm cases are unnecessary, for we can judge cases using intuition alone. In other words, given that intuitions are decisive in decision procedures, if our intuitions can be justified, then it is not clear how decision procedures improve on using intuition alone to decide cases. Conversely, if our intuitions cannot be justified, then it is not clear how decision procedures lead to justified moral beliefs about cases. In recent years many ethicists have defended principlism, casuistry, specificationism, or some other decision procedure on the grounds that it does improve on using intuition, either by eliminating the need for intuition or by minimizing and constraining that need (thus blunting the first horn of the dilemma). I call this the domestication strategy ; the most familiar version of it involves discursive justification. The proposal is that when we reach one of the junctures of judgment, we need not simply assert an intuition and say It seems so to me, but can instead offer reasons for our conclusion about, for example, why we should resolve a conflict between two principles in a particular case in a given way. Such reasons typically consist of philosophical arguments and background theories about the point of the various principles or the moral significance of various features of the case, or of arguments that this way of resolving that

INTUITION AND THE JUNCTURES OF JUDGMENT 3 conflict achieves more coherence among our total set of beliefs (reflective equilibrium is often invoked). I believe that the widespread worry about intuition in moral decision procedures and the drive to avoid it through discursive justification and similar means is misplaced. Unlike other writers on these issues, I will not advocate a particular moral decision procedure or recommend altering any of them. I believe all the major ones already work reasonably well. Instead, I will propose a different understanding of how intuition works in moral decision procedures, and argue for the following claims: A. Discursive justification and other versions of the domestication strategy do not eliminate, minimize, or otherwise constrain the need for intuition. B. However, worries about intuition are overstated; in easy cases we have justified intuitions. C. In harder cases, decision procedures improve on intuition, not by eliminating or minimizing it, but by providing indirect evidence of what our intuition would be were we to see the case under better epistemic conditions where that case is easy and the intuition is be justified. In short, decision procedures function as heuristic devices for helping us reach conclusions that approximate the intuitions we would reach about cases under ideal epistemic conditions our ideal intuitions. D. When decision procedures are simpler, they provide less guidance, requiring us to rely in part on unaided intuition. When they are more elaborate, they provide more guidance for our intuition. Hence, those who distrust intuition in hard cases propose discursive justification and other elaborations to our decision procedures. This makes sense when formulating policies for classes of cases, or in academic discussion. E. However, under clinical conditions, a simpler decision procedure is more epistemically effective even if the judgments it produces are individually less epistemically justified, for given that we cannot avoid acting, we are more knowledgeable overall if our beliefs in all our cases are somewhat justified, than if our moral beliefs about a few cases are very well justified but we simply shoot from the hip in all the others. Moral knowledge differs from other kinds of knowledge in this respect: when I do not need to act on what I know, I can avoid mistake by avowing ignorance on topics where I lack well justified beliefs, but when (as in clinical

4 ethics) I need to act whether I know anything or not, I cannot avoid mistaken action simply by avoiding mistaken belief. F. Thus, discursive justification is entirely appropriate in some circumstances, but seeking it is often epistemically counterproductive at the clinical level, where less is more. G. Once we understand how principles and paradigm cases are derived, we can clarify the truth-conditions for intuitions about where and how indeterminate principles apply, which principle is more weighty when they conflict, which case description captures its morally relevant features, and which paradigm case is relevant similar to the instant case. THE PROBLEMATIC ROLE OF INTUITION IN MORAL DECISION PROCEDURES In this section, I will explain what moral decision procedures are, show where they require intuition, define intuition, and explain why intuitions are often considered epistemically unjustified. Moral theories usually concern act-types, that is, kinds of actions or cases. Act-types can be very general, such as abortion, or very specific, such as aborting an anencephalic to avoid a C-section which would otherwise be contraindicated, where the parents cannot pay for either procedure and the abortion is less expensive, etc. Bioethicists deal in act-types when they deliberate about classes of cases for example, when formulating policies or laws. However, when bioethicists consider particular, actual cases, they are usually dealing in act-tokens. An act-token is a particular instance of an act-type; for example, a particular abortion of a particular anencephalic to avoid a C-section in a particular woman. 1 Act-tokens are actual or possible acts with regard to some actual situation (this actual woman, who may or may not abort this actual fetus). 2 I will sometimes refer to act-tokens as cases, and the case being decided as the instant case. For our purposes, the significance of this distinction is that acttokens must usually be decided in a relatively short period of time by nonphilosophers. 3 Often the right thing to do in such cases is obvious, and can be determined without using moral theory or principles, but sometimes we need some guidance. One can use a moral theory (understood as an explanatory model of our best moral judgments, usually in the form of an organized set of principles) to decide

INTUITION AND THE JUNCTURES OF JUDGMENT 5 whether a given act-token is morally required or prohibited, but moral theories are notoriously ungainly for decisions in real-time by nonspecialists. Most clinical ethicists prefer to use simpler heuristic devices I call moral decision procedures. By decision procedure I mean a method or process 4 for reaching decisions about act-tokens 5 which increases the odds that one s decision will be correct in whatever way moral beliefs can be correct. (Sometimes people use decision procedure to refer to an algorithmic process that resolves all disagreement and requires no intuition; I think this is impossible, and mean the term more broadly.) Popular decision procedures include principlism, the four-topics method, 6 casuistry, specification, common morality theory, 7 narrative ethics, particularist approaches, and others. 8 I will focus on decision procedures that use principles (using this term broadly to include all moral norms) and/or paradigm cases. Principlism 9 and the common morality approach use small sets of simple principles, while the four-topics method uses four small sets of specific principles. Casuistry works by seeking analogies between the instant case and paradigm cases where we feel very confident of what should be done, together with maxims (a kind of small-scale, very specific principle). I will not focus on decision procedures which do not involve principles or paradigm cases, not because I do not recognize their importance, but only because they do not seem to face the issues I want to address. Although my concerns also arise when we apply moral theories to act-types (or act-tokens, for that matter), the proper response to those concerns is different for types and tokens, as I will explain later. Applying principles and comparing paradigm cases requires intuition. What I call intuition goes by several names: judgment, practical wisdom or phronesis, and perception are the most common. In epistemology, an intuition is a belief that is epistemically justified without being inferred from anything else, nor evident through the five senses, introspection, memory, or testimony, and which is self-evident, meaning that if you understand the content of the belief, you are justified in believing it, and if you believe it on the basis of that understanding, you know it to be true. Intuitionism is often dismissed because it is mistakenly thought to require implausibly strong claims: intuitions are infallible and indefeasibly justified (if you have an intuition, you cannot be wrong and you know you cannot be wrong), and we know moral truths intuitively using a special extrasensory faculty for detecting

6 nonnatural properties. However, almost no moral intuitions are that certain, and we cannot account for such a strange faculty or its associated properties within natural science. Fortunately, modern intuitionists are not committed to these claims. Robert Audi, for example, holds that moral intuitions can be wrong, that we can be mistaken about whether they are justified, that we need not know they are self-evident in order for them to be self-evident, that although they can be known noninferentially there may be some intuitions whose content can be also be inferred, and that we apprehend the content of an intuition using our general capacity for reflection and reasoning, rather than a special faculty for sensing weird nonnatural properties. 10 My discussion is premised on this weaker view of intuitions. Let us say that a moral intuition is a belief about some moral issue that is not inferred from anything else, nor acquired through the five senses, introspection, memory, or testimony, and whose alleged justification is that it seems true to the person who has it, and that moral intuition is the ability to acquire moral beliefs justified in that way. 11 Philosophers are accustomed to speaking of intuition a) in connection with our perception of moral principles, as when Ross says we intuitively discern the prima facie duties, and b) in connection with deciding what to do about a case, as when we reach what Rawls called considered moral judgments about cases. However, I am primarily concerned with the third way decision procedures use intuition: c) to ascertain, at what I will call the junctures of judgment, when and precisely how principles and paradigm cases apply or become relevant. The first juncture concerns indeterminacy. A principle is indeterminate with respect to a given case when the principle does not clearly indicate how we must act in that case. For example, we may feel sure we should respect a patient s autonomy, but not be sure whether that means respecting his current competent treatment refusal, or disregarding it on the grounds that we thereby best respect the way he has determined his life overall and over time. We must then use intuition or judgment to determine precisely when and how the principle applies. The second juncture is by far the most familiar: we use intuition to determine which principle to follow when they seem to conflict a moral dilemma. For example, we may have to balance our duty to an existing patient with a duty to take in a new, more seriously ill patient at the cost of neglecting our current patient. The usual way of putting things is that we use intuition to balance the

INTUITION AND THE JUNCTURES OF JUDGMENT 7 apparently conflicting principles and see which has more weight in the case. 12 Third, whether and how a principle applies can vary with how a case is described and understood. A case can be described in many ways, and some case descriptions have different moral implications than others. For example, when describing a case, we decide whether it is relevant that the dying patient s family has neglected her for the last thirty years while her lover has stood by her, and hence whether it is a case of family guilt and conflict over end-of-life care, or a case of how best to respect her precedent autonomy. This too requires intuition. The fourth juncture arises for casuistry, where we judge whether the instant case is sufficiently similar to some paradigm case to require being handled in the same way, and whether the instant case is more like this paradigm case or that one. For example, many of us find Dax Cowart s case a paradigm example of why we should respect a patient s treatment refusal even when we think that means choosing death against his long-range interests. However, it takes judgment to ascertain whether Dax s case is relevantly similar to, say, the case of Terry Schiavo, where keeping her alive went against her earlier wishes, but caused her no pain or suffering. The fact that moral decision procedures are shot through with intuition worries many ethicists, for there are several reasons to question whether moral intuitions are epistemically justified. First, if intuitions were reliable apprehensions of moral truth, we should have the same intuitions more often than we do, yet we often disagree in our moral judgments (and much more often than we do about most empirical questions). Second, as noted earlier, it is hard to characterize intuition as a perceptual faculty similar to vision or hearing, except metaphorically. Third, many intuitions seem clearly wrongheaded, the product of unacknowledged biases or unsupported belief systems. Some of these concerns are blunted but not eliminated once we conceive of intuitions as fallible and only defeasibly justified, and as involving our general capacity for thinking rather than an extrasensory faculty. However, the main objection is that simply seeming true to the believer does not provide sufficient epistemic justification in other words, our supposed intuitions are not self-evident, and all we can do when someone fails to see things as we do is to simply insist that they fail to see it. I see the fourth objection as based on the other three, especially the first two: if we all had the same intuitions, or could scientifically describe a reliable intuitive faculty, we might be less troubled by the noninferential and nonsensory nature of such justification merely seeming true might

8 be enough. 13 Thus, even if we have genuine intuitions of moral truth, we may not be able to tell with sufficient certainty when we have them, and thus we are not justified in believing them. It is tempting to say, Fine, there is some weakness at the joints, but the structure is solid, for so long as our principles and paradigms are well-chosen and carefully applied, their epistemic justification makes up for the lack of justification for our intuitions (something like a well-run civil service where corruption is kept to a workable minimum). However, this response fails, for even at the junctures, intuition is decisive. Consider a conflict of principles, where we must either respect autonomy by honoring the request of an infant patient s parents or follow the principle of justice by denying their request in order to conserve resources for other patients. The difference between judging the first or the second principle to have more weight may mean life or death to one or more patients. The same can be said for intuitions involved in arguments about which features of a case are morally relevant and properly included in the case description, or whether the instant case is more similar to this paradigm case or that one. The looseness at the junctures leaves room for two equally thoughtful deliberators to reach radically different conclusions about the same case. If intuitions sometimes differ, make such a difference, and cannot be justified, then it is hard to see how decision procedures can yield justified moral beliefs about act-tokens. DISCURSIVE JUSTIFICATION Many ethicists are troubled by this. Beauchamp and Childress worry that balancing conflicting principles may be too intuitive and open-ended, 14 that principles are too indeterminate to eventuate in judgments, that without tighter controls on permissible balancing [there will be] too much room... for judgments that are unprincipled, and that we need to reduce intuition to an acceptable level. 15 Henry Richardson claims that [b]y whatever name the intuitive faculty is graced, its operations seem intrinsically beyond the pale of justification. There is no faculty of ethical intuition (or perception or judgment) whose discursively unjustifiable deliverances carry their warrant with them... what it does do remains mysterious or, at best, subjective. 16 Michael Quante and Andreas Vieth contend that unqualified intuitionism does not provide justified moral beliefs. 17

INTUITION AND THE JUNCTURES OF JUDGMENT 9 Over the last fifteen years or so, many ethicists have proposed ways to domesticate intuition, if not by eliminating it from moral deliberation, at least by minimizing the need for it. The most popular domestication strategy is to argue that one s decision procedure can provide discursive justification rather than relying on intuition. Henry Richardson was the first, 18 arguing that his account of how we specify norms when they conflict provides discursive justification, for it does not rely on private and nondiscursive perception or intuition [but on] grounds open to rational public debate and assessment, such as... arguments resting on... underlying theory. 19 David DeGrazia proposed incorporating specification into principlism precisely on the grounds that doing so provides discursive justification; 20 Beauchamp and Childress followed his advice. 21 However, discursive justification need not involve specifying norms; any way of providing articulable reasons and arguments about when and how principles or paradigms apply can be said to provide discursive justification. Carson Strong, for example, contends that casuistry can provide discursive justification through arguments about why cases are decided as they are, thus avoiding intuitions... being used excessively. 22 Gregory Kaebnick says casuists provide it by insisting on consistency between our case judgments, abstracting those judgments into principles, seeking coherence among the principles, and interpreting and clarifying the principles with argumentation (intuitions still play a role, but they receive justification in turn through coherence with moral theory). 23 To understand discursive justification, contrast it with the way intuitions are justified. Intuitions are said to be self-evident; they are not justified by inference from something else. For example, if you assert that a principle applies in a given way to a given case, and that assertion is presented as an intuition, then the intuition is supposed to be justified by the fact that comprehending the content of the intuition justifies you in believing it. Discursive justification, by contrast, involves justifying an assertion by inferring it from or supporting it by something else typically arguments, reasons, or interpretations. 24 The relevant meaning of discursive is [p]assing from premises to conclusions; proceeding by reasoning or argument; ratiocinative... Often opp. to intuitive. 25 Thus, you discursively justify your assertion that the principle applies in a given way by offering arguments and reasons for that assertion, rather than treating the assertion as evident on its face. One of Richardson s examples of discursive justification involves constructing arguments and developing reasons for

10 the prohibition on killing by tracing it back to a more general principle of respect for self-conscious life: [A]lthough each of these specifications is controversial, their grounds are not some kind of private and nondiscursive perception or intuition. Rather, they rest on grounds open to rational public debate and assessment, such as those arguments resting on the underlying theory of respect for persons. 26 Perhaps in hopes of avoiding the need for self-evident foundational beliefs, discursive justification is often said to be strongly coherentist, 27 such that principles are both consistent with each other and supply argumentative support for each other. Richardson often speaks of reflective equilibrium, and declares his account to be coherentist. 28 Similarly, DeGrazia says a norm specification is discursively justified when it maintains or increases the coherence of our entire set of norms, not only by avoiding contradictions among them, but by the fact that the norms and the reasons behind them are mutually supporting. 29 Nor is discursive justification the only domestication strategy. Beauchamp and Childress incorporate specification and hence discursive justification into principlism, but they also believe that proper safeguards can be put in place that will adequately manage the problems, 30 such as six limitations on the use of intuition in resolving conflicts of principles (for example, when we violate one principle in order to follow another, we must violate it in the least intrusive way, and we must act impartially and not be influenced by morally irrelevant information about any party). 31 Unfortunately, these domestication strategies are unsatisfactory for three reasons. First, there is a large difference, somewhat glossed over in the discussions cited above, between being able to provide discursive justification and actually providing it. Proponents of these strategies tend to say that principlism, specification, or casuistry can be supplemented with discursive justification, not that they always are. However, you are not justified in believing something simply because it is possible for you to have evidence or a reliablyfunctioning perceptual process you are justified only if you actually have those things. Moreover, providing discursive justification is not easy, and may require considerable time and philosophical sophistication. Decision procedures, however, are designed for use by nonphilosophers under real world conditions with insufficient time for philosophical labor and moral theory construction. Thus, in practice moral decision procedures may lack discursive

INTUITION AND THE JUNCTURES OF JUDGMENT 11 justification. If we could be sure that such justification is always available for any given intuition at the junctures of judgment, then we would have a metajustification for such intuitions, and would not require discursive justification. However, discursive justification is thought to be needed precisely because there may be trouble at the junctures. Second, whatever problems arise with intuition persist even when a decision procedure provides discursive justification. No matter how carefully we analyze and construct arguments, draw inferences, or make interpretations, we use intuition to assess whether the inferences are strong, the arguments cogent, and the interpretations insightful and correct. Outside of formal logic, these questions are matters of judgment for which there is no algorithm. For example, Richardson describes someone deliberating about whether to withhold nutrition and hydration from a severely deformed newborn to let it die; this deliberator considers the reasons behind the general prohibition on killing, concludes that it has to do with the value of persons, and decides, on reflection, that the value of persons stems from the value of self-conscious life or the potential for such life. 32 How can these issues be settled without intuition? If arguments are made, then premises must be established and inferences defended again by intuition. It is precisely because we disagree about such intuitions that arguments often fail to produce agreement. Even if the structure of justification is entirely coherentist, with no beliefs playing a foundational role, we must still use intuition to assess which coherent set of moral beliefs is superior (for there is always more than one possible coherent set) and still use intuition to assess the strength and cogency of the arguments in the coherent set. Third, elaborating a decision procedure with arguments and a search for coherence may multiply the junctures at which intuition is needed and potentially decisive. Consider the following (actual) case. A competent, elderly patient who has never been a Jehovah s Witness refuses blood transfusions for his upcoming heart surgery because a few days ago a Witness relative showed him a TV documentary about the dangers of blood transfusions; his surgeons say the surgery will be extremely risky without transfusions. Now suppose we discursively reason as follows: we decide (using intuition) that the point of respecting a patient s autonomy is to respect the patient s efforts to determine his own life. This is partly because (we intuit that) self-determination has intrinsic moral value, and

12 because (we intuit that) people tend to be the best judges of their own interests. We thus conclude (by intuition) that a life lived in accord with long-term beliefs and intentions is more self-determined than a life that departs from such beliefs and intentions. This, however, seems inconsistent with respecting changes of mind, and we decide (by intuition) that only a carefully considered and wellinformed change of mind can override a long-term belief or value, especially when great risks are at stake (which we gauge using intuition). Finally, we decide (by intuition) that this patient has not made a sufficiently careful and well-informed change of mind, and (we intuit that) the risks are too great. One may object that I have overstated the occasions requiring intuition here; for example, one can argue for the value of selfdetermination rather than simply intuit it, or make empirical observations that people tend to be happier when their self-determination is respected. 33 (This is consistent with the view of intuition offered earlier: just because some moral fact can be intuited does not mean that it can be known only through intuition.) However, if we argue for this claim, we then further multiply the places at which intuition is needed, for some of the premises of that argument will have to be intuited, and we must intuitively apprehend that certain conclusions follow from those premises, and so on. If, by contrast, we offer evidence that respecting self-determination tends to promote an agent s well-being, then we need to intuit how well-off the agent is or she must intuit that for herself and tell us. Even if well-being is defined as something we can ascertain without intuition, such as preference satisfaction, the very claim that this is what constitutes well-being requires intuition to establish, either directly or by argument and further intuitions. In any case, even if some of the matters intuited above can be ascertained entirely without intuition, not all of them can. In our search for discursive justification, we used intuition several times in this course of reasoning, and even though many of these issues may seem narrow, at least some of these intuitions are decisive (had they gone the other way, we might have decided the case differently). If we used intuition to simply apply the principle of respect for autonomy without further argument, we could use intuition just once. Thus, the more discursive justification we build into a decision procedure, the more often we must exercise intuition decisively.

INTUITION AND THE JUNCTURES OF JUDGMENT 13 MAKING HARD CASES EASY AS A WAY OF HAVING JUSTIFIED INTUITIONS Intuition cannot be domesticated, but fortunately the problem is overstated. It helps to distinguish between philosophical skepticism and practical skepticism about intuition. For illustration, consider philosophical skepticism and practical skepticism about perception. We have philosophical reasons to wonder whether any of our perceptions are epistemically justified; perhaps we are victims of a Cartesian evil demon. However, we trust most perceptions in practice; practical skepticism arises only when we have some reason to distrust a particular perception or class of perceptions. For example, perhaps eyewitnesses to a shooting are too overwhelmed to have reliable perceptions of the event, or perhaps most of us lack the perceptual skill to make fine discriminations about color or taste. No matter how intractable philosophical skepticism about perception may be, that kind of skepticism is no reason to distrust all perceptions in practice. The worries about intuition mentioned earlier (disagreement, no scientific model for moral perception, and possible bias) support philosophical skepticism about intuition; they are reasons to wonder whether any intuitions are epistemically justified. However, no philosophical skeptics distrust all intuitions in practice. Bias can be counteracted, and there is a range of easy cases where we tend to agree and feel confident of our moral intuition. As for lacking a scientific model of a faculty of moral intuition akin to our models of vision or hearing, we also lack such a model for intuition about arithmetic or simple logic, but no one doubts our intuitions about the validity of modus ponens, or the truth of 5 + 7 = 12. When assessing moral decision procedures, it is no more reasonable to raise epistemic concerns that apply to all of ethics than it is to raise skeptical concerns about perception generally when assessing telescopes for accuracy. Just as we must accept many perceptions in practice, so we must accept many moral intuitions in practice. The literature on moral decision procedures is distorted by a tendency to argue as if a procedure had to resolve all our concerns about intuition. A moral decision procedure must pass only a more modest test: can it yield moral intuitions about act-tokens which are sufficiently similar to intuitions we regard as sufficiently justified for practical purposes?

14 I will argue that easy cases can be understood in terms of favorable epistemic conditions, and that intuitions formed under ideal epistemic conditions would all involve easy cases and hence be justified. I will then show how decision procedures can yield justified moral beliefs under nonideal conditions. Here is the overall argument: 1. All moral facts are knowable in principle. 2. Our moral intuitions in easy cases are justified. 3. Cases are easy when the epistemic conditions are suitably favorable. 4. Therefore, all hard cases are potentially easy cases. (1, 3) 5. Therefore, it is possible to have justified intuitions in all cases, including the hard ones. (2, 4) Premise 1: All moral facts are knowable in principle. This is not to say that all moral truths are knowable as a practical matter; some may be impossible to discern under any realistic conditions. However, in principle under ideal epistemic conditions, however distant those are from any actual epistemic conditions we could know them. It is hard to imagine how moral truths could not be knowable in principle (the only possibility I know of combines two claims: the divine command theory of ethics is true God determines what is right and wrong and we could not comprehend God s mind even if we had access to it). I will assume that most readers find it plausible that moral truth is always knowable in principle. Premise 2: Intuitions in easy cases are justified. Again, the argument for this premise is not that we can overcome all skeptical doubts about moral intuition; the argument is that we do not need to. Premise three: Cases are easy when the epistemic conditions are suitably favorable. Easy case is defined in terms of ease of discerning moral truth. For example, it is clear and undisputed that we should strive to save an ailing infant with a prospect of a long and healthy life where this will not risk harm to anyone else. This is an easy case not because it is easy to do this, but because it is easy to see that we should do this. A case becomes easy when the epistemic conditions are favorable enough for us to easily see what should be done. Those conditions may vary with the moral truth in question, or with the skill of the moral observer. For example, it is a bit

INTUITION AND THE JUNCTURES OF JUDGMENT 15 harder to see that we should save an ailing infant where doing so takes scarce medical resources away from another ailing infant with lesser prospects of a long and healthy life, but it might be easy to see the truth about that case if we have more time to deliberate, more information, greater familiarity with relevant moral background theory, a greater capacity for empathetic awareness, and so on. The second case becomes easy under the right epistemic conditions. Step four, first conclusion: Therefore, all hard cases are potentially easy cases. This follows from steps one and three. To put this another way, there are no cases so hard that the moral truth is unknowable under all conceivable conditions. Step five, second conclusion: Therefore, it is possible to have justified intuitions in all cases, including the hard ones. This follows from steps two and four. Again, having such an intuition is not a realistic possibility in all cases, but it is possible under some ideal epistemic conditions (which may never occur). This argument shows that, so long as we accept intuitions in easy cases, we should assume that justified intuitions are possible in all cases. Another way to put this is to say that, for every moral question, there is a corresponding ideal intuition we would reach if the epistemic conditions were sufficiently favorable to make the case an easy one. Therefore, when we make judgments about act-tokens, we want to know that our actual intuitions are reasonably likely to approximate their corresponding ideal intuitions. This means that moral intuitions can be justified in two ways. First, they can be self-evident, as they are in easy cases. This is the way most philosophers have thought intuitions had to be justified. However, to say that a proposition is self-evident is to say that it can be self-justifying, not that it can be justified only by being self-evident. There is no reason why some self-evident propositions cannot be justified in more than one way, both by being self-evident and by being inferred from something else. Second, moral intuitions can be justified by evidence that they coincide with their corresponding ideal intuitions. This involves self-evidence only in the derivative sense that the ideal intuition which may be only hypothetical is self-evident. Rather, it requires arguments and inferences to show that the actual judgment is likely to approximate its corresponding ideal judgment. Actual intuitions in easy cases will be justified by self-evidence, but actual intuitions in hard cases will have to be justified in the second

16 way, which I will call indirect justification. This is where decision procedures come in. PRINCIPLES AND PARADIGM CASES AS HEURISTIC DEVICES This paper began with a dilemma about the relationship between principles and cases: if intuitions are justified, then it is not clear why principles and decision procedures are ever necessary, while if our intuitions are not justified, then (given that intuitions are decisive) it is not clear that principles and decision procedures are ever sufficient for reaching justified moral beliefs about cases. The first step toward avoiding the dilemma is to show that some intuitions are justified the ones we have about easy cases. We also saw that similarly justified intuitions are possible (in principle) in all hard cases. But how do we turn a hard case into an easy case? To answer this, we must understand where principles and paradigm cases come from in the first place. Suppose we have a clear, confident intuition about some acttoken an easy case and we want to summarize that intuition for future use or communication to others. We can take an intuition in a single easy case and abstract from that case, deriving a moral principle. We can also take a number of easy cases that exhibit a pattern and derive a principle from them all. We can also assemble these principles into a moral theory. It is an old idea: Mill spoke of the rules of common sense morality as rules of thumb which help us approximate the moral judgments a perfect utility-calculator would reach; more recently, Rawls has floated a nonutilitarian version of this idea. 34 Principles, Rawls argues, are devices to help us reach the intuitions we would reach if we were fully competent moral judges forming intuitions under the right conditions. We are not always in that position, but we can abstract principles from our best intuitions, and use those principles to infer what our best intuitions are likely to be in relevantly similar situations where our epistemic circumstances are not favorable enough to enable us to reach such intuitions directly. Because principles and paradigm cases are derived from easy cases, they can serve as heuristic devices to help us reach actual intuitions about act-tokens which are likely to coincide with or at least approximate their corresponding ideal intuitions. This is a way

INTUITION AND THE JUNCTURES OF JUDGMENT 17 of providing indirect justification for our intuitions in hard cases somewhat like the indirect justification you get when you know that you see a case the way some better, more discerning moral judge sees it. For example, if under suitably favorable epistemic conditions we would decide to go along with a particular patient s request even though his request imposed a mild risk upon his family, then applying the principle of respect for autonomy and the principle of justice and balancing the two might help us to focus on the relevant features of a case and isolate them from distracting and irrelevant features, so that it is easier to intuit what should be done. Principles can even arrange the issues so that those with implications for other issues get judged first (ascertain the patient s competence before deciding how to balance respect for his autonomy with concern for his family members). When we derive a principle, we can do so at various levels of generality. Think of a spectrum. At the concrete end is a highly detailed and specific statement of our intuition, including both some normative content about what should or should not be done, and an act-description with all the features of the token we found morally relevant (that is, features whose inclusion or exclusion from the acttoken alters our intuition). If we add a maxim to the statement, these become the paradigm cases the casuists speak of; this is why I often speak of principles and paradigm cases in the same way from here on. This also explains why casuistry seems compatible with principlism: the difference between paradigm cases and principles is one of degree they both lie on such spectra. 35 At the other end of the spectrum is the most abstract principle one can derive (something like Do the right thing ). Between these extremes there are successively more or less abstract descriptions of the case and its normative content. The more abstract the description, the more it looks like a principle and not a case description with normative content. I call this the principle-case spectrum, and will refer to everything on it as a normative description. There is a spectrum for every case, for in every case it is possible, under ideal conditions, to have a justified intuition and derive a principle from it. It seems likely that these spectra tend to converge at the more abstract end (something like a teepee with branching poles), eventuating in a small set of principles (or, in a monistic system, just one). In any event, a principle applies to a case to the extent it matches the principle we would derive from that case under ideal conditions (even if the principles themselves are different). 36

18 If all goes well, when we apply an applicable principle, we get a judgment about the case that matches the judgment we would reach if we used intuition alone under ideal conditions the principle guides our intuition. The problem that vexes proponents of decision procedures is that principles do not always provide as much guidance as we would like in hard cases, leaving it unclear what judgment to reach. Principles drawn from the more abstract end of a given principle-case spectrum work well when we need only a little heuristic guidance in order to reach something approximating an ideal judgment, and sometimes the right judgment is so obvious that we do not need heuristics at all. However, in harder cases we must use intuition when it is not immediately clear where and how a general, abstract principle applies. At the point where guidance gives out, we must rely on unguided intuition. More abstract principles put us at that point sooner. Thus, we may want to trace our way back to the more concrete, specific end of the spectra in order to find a more detailed principle or paradigm case that gives our intuition more guidance. This accounts for the tendency of some ethicists to elaborate decision procedures in various ways: by adding subprinciples, or by splicing principlism and casuistry and wide reflective equilibrium and specificationism together or seeking discursive justification through further moral theorizing, elaborating and arguing for principles and their relationships to each other and to cases. These addenda give our intuitions more guidance in hard cases. In other words, if your principle is too abstract for the case, then instead of using intuition to fill the gap, as it were, engage in moral theory construction by developing a more articulate principle that provides more detailed guidance. However, where moral decision procedures are concerned, this is the wrong way to deal with hard cases, for it can actually decrease our moral knowledge overall. Unlike moral theories, moral decision procedures are meant for nonphilosophers deciding act-tokens under real-time conditions. Such decision procedures must increase our knowledge in two ways that often conflict: they must provide guidance (e.g., must maximize our odds of reaching an intuition about an act-token that reasonably approximates our ideal intuition about that token), and they must be user-friendly (e.g., they must be sufficiently easy to use that nonspecialists will put them to work in real life).

INTUITION AND THE JUNCTURES OF JUDGMENT 19 There are different ways to strike a balance between guidance and user-friendliness. To take a simple example, we might have either a smaller set of more abstract principles, or a larger set of more detailed principles. A larger set of more detailed principles provides more guidance, for each principle provides a more detailed act-description to tell us what it applies to, but a larger set of more detailed principles is also less user-friendly than a smaller set, for there is more information to remember and apply. A smaller set of more abstract and general principles will be easier to use and remember, and hence more user-friendly, but because each principle is less specific, and there are fewer of them, the odds that any one of them will clearly resemble the instant case are less, and hence they provide less guidance. Thus, a larger set of more specific principles provides greater guidance at the expense of user-friendliness, while a smaller set of more abstract principles is more user-friendly, but at the cost of providing less guidance. To see why seeking guidance alone is mistaken for decision procedures, let us distinguish decision procedures from moral theories. By moral theory I mean a representation or model of our moral beliefs as a set of principles, usually by abstracting the principles from patterns in our best moral intuitions. 37 (Some authors impose much stricter requirements on what counts as a moral theory; stricter definitions do not conflict with my argument.) Moral theories and decision procedures are similar in that most of them use principles of some kind, and this similarity can be misleading. A moral theory is not necessarily a decision procedure, though many moral theories can be used as one, and a decision procedure need not use a moral theory (or part of one), though it might. They are distinguished primarily by their purposes. Decision procedures can be used on act-types, but typically they are used on act-tokens that must be decided by nonphilosophers working under severe real-time constraints. Moral theories, by contrast, are meant to accurately and completely model our best moral judgments, and are not meant to be used to decide act-tokens under practical realtime constraints. 38 We should understand a moral theory as a representation of our moral beliefs which maximizes guidance without regard to loss of user-friendliness. This fits our general notion of theories as representations or models of reality, where accuracy and detail are prime virtues. However, maximizing guidance without regard to user-friendliness is a bad strategy for developing an effective decision procedure.

20 Thus, the strategy of elaborating our decision procedures to provide more guidance is epistemically costly, for if a decision procedure is too cumbersome, it simply will not be used at all, or at least the elaborated parts of it will not be used. Often we are better advised to use a simple decision procedure (the four principles, for example, without any elaborations or attempts at specification or achieving reflective equilibrium, or any mention of sub-principles and background theories behind the principles), and rely on our unaided intuition even if we are not entirely confident of it. Moral decisions must be made in real-time by people not specialized in moral theory, and a decision procedure must be fairly simple, or it will not be used at all. Moreover, even if it is used, the time and effort involved in using it may make it difficult to address all ethics cases in detail; it could turn out to be the ethics equivalent of a gilded coach wheeled out on ceremonial occasions but useless for the daily commute. To illustrate, suppose you can either have a very well justified moral intuition about 10 cases over the next year, or less well-justified intuitions about 100 cases over that year. Which outcome leaves you better informed? Under the latter scenario, each decision in the 100 cases will be less well justified, but under the former, 90 of that 100 will be even less justified still. If we could simply take on a smaller workload (like an appellate court restricting the number of appeals we can hear), that might not matter so much, but in fact each of those 100 cases will be decided in any event, for doing nothing is tantamount to a decision too. If we want to maximize our moral knowledge in a morally responsible way, we should not seek the most justified intuitions we can have (accepting that this means having fewer justified beliefs), but should instead strive to achieve the highest level of justification we can for our intuitions in all the cases we decide. Why is this not noted more often? I suspect two reasons. First, it is easy to overlook the distinction between philosophical skepticism and practical skepticism. We make the worries about hard cases seem more intractable than they really are when we write as though resolving them requires resolving philosophical skepticism about intuition. Second, it is easy to think that how knowledgeable we are about moral matters turns directly on how justified our moral intuitions are. However, being knowledgeable about moral matters is more complicated than that, for maximizing justification may leave us with a few exquisitely justified moral beliefs amid large