The Voice of Conscience

The Voice of Conscience Author(s): J. David Velleman Reviewed work(s): Source: Proceedings of the Aristotelian Society, New Series, Vol. 99 (1999), pp. 57-76 Published by: Blackwell Publishing on behalf of The Aristotelian Society Stable URL: http://www.jstor.org/stable/4545295 . Accessed: 03/01/2012 10:48 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. The Aristotelian Society and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the Aristotelian Society. http://www.jstor.org IV*-THE VOICEOF CONSCIENCE1 by J. DavidVelleman I reconstructKant'sderivationof the CategoricalImperative(CI) as an argumentthatdeduces whatthe voice of conscience must say from how it must sound-that is, from the authoritythatis metaphoricallyattributedto conscience in the form of a resoundingvoice. The idea of imagining the CI as the voice of conscience comes from Freud;and the presentreconstructionis partof a larger project that aims to reconcile Kant's moral psychology with Freud's theory of moral development. As I reconstructit, Kant's argumentyields an imperative commandingus to act for reasons whose validity we can consistently will to be common knowledge among all agents. Universalizinga maxim thus turnsout to consist in willing, not that there be some universallyquantifiedrule of conduct, but ratherthat a principle of practicalreasoning be common knowledge-as a principleof reasoningought to be. ABSTRACT Howdo you recognizethe voice of your conscience?One possibility is that you recognize this voice by what it talks about-namely, your moral obligations, what you morally ought or ought not to do. Yet if the dictates of conscience were recognizableby theirsubjectmatter,you wouldn'tneed to thinkof them as issuing from a distinctfacultyor in a distinctivevoice. You wouldn'tneed the concept of a conscience, any morethanyou need concepts of distinct mental faculties for politics or etiquette.Talk of conscience and its dictates would be like talk of the mince-pie syllogism, in that it would needlessly elevate a definable subject matterto the status of a form or faculty of reasoning.2 1. In writing this paper I have drawn on conversationsand correspondencewith Marcia Baron,JenniferChurch,Stephen Darwall,David Hills, David Phillips, and Connie Rosati. Workon this paperhas been supportedby a sabbaticalleave fromthe College of Literature, Science, and the Arts, University of Michigan; and by a fellowship from the National Endowmentfor the Humanities. 2. The mince pie syllogism was the ironic invention of Elizabeth Anscombe. Anscombe objected to the notion that the practicalsyllogism was merely a syllogism on a practical topic, such as what one ought to do. She arguedthatif therewere a distinctlogical form for reasoning about what one ought to do, then there might as well be distinct forms for reasoningaboutevery definabletopic, includingmince pies. (Intention[Ithaca,NY: Cornell University Press, 1957], 58.) *Meeting of the Aristotelian Society, held in Senate House, University of London, on Monday, 23rd November, 1998 at 8.15 p.m. 58 J. DAVIDVELLEMAN Ourhavingthe conceptof a conscience suggests, on the contrary, that ordinarypracticalthought does not contain a distinct, moral sense of 'ought' that lends a distinct, moral content to some practical conclusions. The point of talking about the conscience andits voice is precisely to marka distinctionamongthoughtsthat are not initially distinguishable in content. Among the many conclusions we drawaboutwhatwe oughtor oughtnot to do, some but not others resonate in a particularway that marks them as dictates of conscience. The phrase 'morally ought' is a philosophicalcoinage that introducesa differenceof sense where ordinarythoughthas only a differenceof voice-whatever thatis. But what is it? Conscience doesn't literally speak. The idea of its addressing you in a voice is thus an image, albeit an image that may infiltrateyour experience of moralthoughtand not just your descriptions of it. Yet whether the dictates of conscience are somehow experiencedas spoken or arejust describedas such after the fact, this image must represent something significant about them, or it wouldn't be used to identify them as a distinctivemode of thought.The questionis what literalfeatureof these thoughtsis representedby the image of theirbeing deliveredin a voice. The answer, I think, is that the dictates of conscience carry an authoritythat distinguishesthem from other thoughtsabout what you ought or ought not to do.3 The voice of conscience is, metaphoricallyspeaking,the voice of this authority.To recognize an 'ought' as delivered in the voice of conscience is to recognize it as carrying a different degree or kind of authorityfrom the ordinary 'ought', and hence as due a differentdegree or kind of deference. If the voice of conscience does representa distinctiveauthority that accompanies some practicalconclusions, then it is more than a curiosity of moral psychology: it symbolizes a fundamental feature of morality, regarded by some philosophers as the fundamentalfeature.Kant,in particular,thoughtthatwhatmorality requirescan be deduced from the authoritythat must accompany its requirements.If Kanthad writtenin the imageryof conscience, he might have put it like this: by reflecting on how the voice of 3. The authority of conscience is the central theme of Butler's Sermons. For a recent discussion of Butler, see Stephen Darwall, The British Moralists and the Internal 'Ought' 1640-1740 (Cambridge:CambridgeUniversityPress, 1995), Chapter9. THE VOICEOF CONSCIENCE 59 conscience must sound, you can deduce what it must saywhereuponyou will have heardit speak. Of course, Kantdidn'tformulatehis moraltheoryin these terms, but I thinkthatthey can be substitutedfor termssuch as 'duty' and 'moral law' in Kant'sown formulations,with some gain in clarity and persuasivenessfor modem readers.My goal is to reconstruct Kant's categorical imperativein the terms of conscience and its voice.4 The idea of reconstructingthe categoricalimperativeas the voice of conscience originatedwith Freud.Freud was interestedin the voice of conscience because he thoughtthat it could explain why paranoiacsheardvoices commentingon theirbehaviour;5and that it could in turn be explained by the psychological origins of conscience in parentaldiscipline 'conveyed... by the medium of the voice'.6 In tracing conscience to the voice of parental discipline, Freudalso thoughtthathe could explain why its power 'manifests itself in the form of a categorical imperative.'7This explanationshowed, accordingto Freud,that 'Kant's Categorical Imperativeis... the direct heir of the Oedipuscomplex'.8 My view, which I cannot defend here,9 is that the categorical imperativecan indeed be identified with the super-ego,at least in one of its guises. For I thinkthatthe categoricalimperativeis what Freud would call an ego ideal. The ego ideal, in Freudiantheory, is that aspect of the super-egowhich representsthe excellences of parental figures whom the subject loved and consequently 4. There is at least one passage in which Kant uses the word 'conscience' in reference to the activity of applying the categorical imperative: Groundworkof the Metaphysic of Morals, trans.H.J. Paton (New York:Harperand Row, 1964), 89 (422). (Page numbersin parenthesesrefer to the PrussianAcademy Edition.) 5. 'On Narcissism: An Introduction', in The Standard Edition of the Complete Psychological Worksof SigmundFreud, Vol. 14, ed. JamesStrachey(London:The Hogarth Press, 1957), 69-102, at 95. See also GroupPsychology and the Analysis of the Ego, S.E. 18: 67-143, at 110 [53]; New IntroductoryLectureson Psycho-Analysis,S.E. 22: 3-182, at 59 [74]. (Page numbersin bracketsrefer to the Norton paperbackvolumes of individual works from the StandardEdition.) 6. 'On Narcissism', 14: 96. 7. TheEgo and the Id, S.E. 19: 3-66, at 35, 48 [31, 49]. Freudalso uses this phrasein Totem and Taboo, S.E. 13: ix-1 62, at 22. 8. 'The Economic Problemof Masochism', S.E. 19: 156-70, at 167. Freudalso identified the super-ego with the Kantian'moral law within us' (New IntroductoryLectures,22: 61, 163 [77, 202]). 9. But see 'The Direct Heir of the OedipusComplex' (MS). 60 J. DAVIDVELLEMAN idealized when he was a child.10Although Kant often framedthe categoricalimperativeas a rule for the will to follow, I thinkthatit is better understoodas an ideal for the will to emulate, in that it describes an ideal configurationof the will itself. And I think that this ideal could indeedbe internalizedfromparentalfiguresas they appearto the eyes of a loving child. This conceptionof the categoricalimperativeas an ego ideal will reappearat the end of this paper,butit is not my immediateconcern. What concerns me here is Freud'ssuggestion that the categorical imperativecan be identifiedwith the voice of conscience. The image of conscience as having a voice is potentially misleading in one respect. Taken literally, the image may lead us to thinkof conscience as an externalintelligence whisperingin our ears, like Socrates's daimon. Even when taken figuratively, the image still suggests that the dictates of conscience occur to us unbidden,as thoughts that we don't actively think for ourselves, and hence as externalto us, in the sense madefamiliarby the work 11 of HarryFrankfurt. Conscience is most likely to seem externalin this sense when it opposes temptation: conscience and temptation can seem like parties to a dispute on which we sit as independentadjudicators. Yet even this judicial image is misleading, since the disputing partiesdo not appearas distinctfromourselves.We ourselvesplay each role in the mental courtroom,now advocating the case of temptation, now that of conscience, representing each side in propriapersona. In short,we vacillate-which entails speakingin differentvoices, not just hearingthem. Thus, hearing the voice of our conscience is not really a matter of hearing voices. It's rathera matter of recognizing a voice in which we sometimes speak to ourselves. Freud's theoryof the super-egomight seem to favourthe image of conscience as an independent agency, distinct from and in opposition to the self. Freud certainly thought that in cases of 10. Freud's views on the relationbetween super-egoand ego ideal are clearly summarized in Joseph Sandler,Alex Holder,and Dale Meers, 'The Ego Ideal and the Ideal Self', 18 The PsychoanalyticStudyof the Child 139-58 (1963). See also JosephSandler,'On the Concept of the Superego', 15 ThePsychoanalyticStudyof the Child 128-62 (1960). 11. The Importanceof What We Care About (Cambridge:CambridgeUniversity Press, 1988), especially chapters2, 5, 7, and 12. THE VOICEOF CONSCIENCE 61 mental illness, the super-ego could become the source of voices heardinvoluntarily,and hence from outside the self in Frankfurt's sense.12 Yet in the normal subject, the super-ego bears an ambiguousrelationto the self. It is 'a differentiatinggrade in the ego',13 and the process of introjectionby which it is formed is a way of identifying with other people, which is necessarily a deployment of the self. So anotherdescriptionof what happens when the super-egoaddressesthe ego is thatthe self identifies with others in addressingitself. Although Kant doesn't tend to speak of the conscience per se, his moral philosophy also reflects the complexity of its relationto the self. On the one hand,Kantsays thatthe morallaw is necessary and inescapable;on the otherhand,he describesit as a law thatwe give to ourselves. For Kant, giving ourselves the moral law represents both our exercise of an autonomous will and our subjectionto a necessity largerthan ourselves;just as, for Freud, conscience is the ego addressing itself in the voice of external authority.14 This analogy reveals what is right about Freud'sclaim that the voice of the super-egois the voice of Kant'scategoricalimperative. The necessity to which we submit in the law that we give to ourselves can be imaginedas the authoritywe recognize in a voice with which we address ourselves-namely, the voice of conscience. I wantto show thatKant'sattemptto derivethe content of the moral law from the very concept of its practicalnecessity can be restaged as an attemptto derive the words of conscience from the authoritativesound of its voice. An example of rational authority. The first step in this reconstructionof Kantianethics is to analyze the authoritythat Kantwould attributeto the conscience. WhereasFreudthoughtof the conscience as the seat of internalizedparentalauthority,Kant would think of it-if he thoughtin such terms at all-as a seat of rationalauthority.But what sort of authorityis that? 12. See the passages cited in note 5, above. 13. This is the title of ChapterXI of GroupPsychology. 14. Kant seems to reject the image of an externalvoice of conscience at Groundwork93 (425-26), where he insists that moral philosophy cannot serve 'as the mouthpieceof laws whisperedto her by some implantedsense or by who knows what tutelarynature...'. 62 J. DAVIDVELLEMAN Consider, by way of analogy, the authority of cognitive judgments whose propositionalcontentis self-evidentlytrue. You make such a judgment,for example,when you confirmfor yourself that 2+2=4. To say that such a judgment is authoritativeis to say that it merits deference. But why should anyone defer to your judgment on mattersof elementaryarithmetic? The answeris not thatyou'reespecially well positionedto think about such matters.When it comes to adding 2 and 2, all thinkers are in the same position. But for that very reason, a computation performedby you here and now can take the place of anyone's, including your own on futureoccasions. Thatis, you can compute the sum of 2 and 2 once and for all, in that you would only computeit similarlyin the future;andyou can also computeit one for all, in that others would only compute it similarly,too. Your judgment is thus authoritativebecause it can serve as proxy for anyone's, including your later selves'. To see yourself as judging authoritativelyis to see yourself as judging for all in this sensein the sense, that is, of judging as anyone would. But whatif yourjudging as anyonewould were, in turn,a matter on which judgments might differ? In that case, your arithmetic judgment might only seem authoritativeto you. Surely, however, you recognize your judgment as having an authoritythat anyone would recognize. You must thereforesee yourself as judging, not just as anyone would, but as anyone would judge that anyone would. And now an infinite regress rearsits head. For what if judgments could differ as to whetheryou werejudging as anyonewouldjudge that anyone would-and so on? Fortunately,there is independent reason to expect such a regress in the presentcontext, and also to regardit as benign. The reasonis thatthe facts of elementaryarithmeticarecommon knowledge among those who consider them, and common knowledge involves a regress of the present form. Anyone who adds2 and2 sees, notjust thatit's 4, butalso thatanyonewho added 2 and 2 would see thatit's 4, and thatsuch a personwould see this, too, and so on. The facts of elementaryarithmeticare like objects in a public space, whereeveryonesees whatevereveryoneelse sees, and everyone sees everyone else seeing it. Unlike publicly visible objects, however, the facts of arithmeticare common knowledge THEVOICEOFCONSCIENCE 63 among all possible thinkersratherthana finite populationof actual viewers. As a participantin this common knowledge, you have higherorder knowledge about the judgments of all other thinkers, and abouttheirjudgmentsaboutthejudgmentsof all. This higher-order knowledge constitutes a perception of authority in your own judgment that 2+2=4, since it representsthis judgment as that which anyone would think, and would think that anyone would think, and so on. So it's just as we might have expected:the voice of authorityis the one with the reverb. But now we know the source of the reverberations.A judgment resounds with authority when it is perceived as echoing and re-echoing in the minds of all other thinkers, as it does when its content is a matter of common knowledge. This authorityattaches, as we have seen, to items of a priori knowledge, such as the judgment that 2+2=4. Items of a priori knowledge would seem to be the only bearersof this authority,in fact, since only the a priori can be regardedas what anyone would think, or be thoughtto think, and so on. The authorityof the moral law. I suspect thatthe form of common knowledge amongall thinkers-of thatwhich anyonewould think, and would think that anyone would think, and so on-is the form that Kant attributesto the moral law in calling it universal. Of course, Kantthinksthatthe morallaw is universalin the sense that it applies to all rationalcreatures;andthe most economical way of representing a universally applicable law is with a universal quantifier,as in 'All rationalcreaturesmust keep their promises' or 'No rational creature may lie'. But serious problems, both textual and philosophical, stand in the way of readingKant's talk of universallaw as referringto universallyquantifiedrules. Consider, to begin with, these two passages from the Groundwork. Everyone must admitthata law, if it is to hold morally-that is, as the groundof an obligation-must carrywith it absolutenecessity; that the command 'Thou shalt not lie' does not hold just for men, without otherrationalbeings having to heed it, and similarly with all the othergenuine morallaws; and thatconsequentlythe ground 64 J. DAVIDVELLEMAN of obligationhere mustbe sought, not in the natureof manor in the circumstancesof the world where he is located, but solely a priori in the concepts of purereason.15 It may be added that unless we wish to deny to the concept of morality all truthand all relation to a possible object, we cannot dispute that its law is of such widespreadsignificance as to hold, not merely for men, butfor all rationalbeings as such-not merely subjectto contingentconditionsand exceptions, but with absolute necessity....And how could laws for determiningour will be taken as laws for determiningthe will of a rationalbeing as such-and only because of this for determining ours-if these laws were merely empiricaland did not have theirsource completely a priori in pure,but practicalreason?16 These passages are central to the Groundwork,because they introducethe conceptualconnectionsamongmorality,universality, and the a priori-the connections throughwhich Kant hopes to derive the content of the categorical imperative from the very concept of morality. The passages argue that the concept of morality entails that its laws carry 'absolute necessity'; which entails thatthey hold not only for men butfor all rationalcreatures; which entails thatthey hold a priori. Suppose thatwe interpretthis argumentas using the word 'laws' to denote general rules, and as contrastingrules thatquantifyover men with rules thatquantifyover rationalcreatures.We must then wonderwhy the formerrules are any less necessarythanthe latter, since the formerapplynecessarilyto anythinginsofaras it is a man, just as the latter apply necessarily to anything insofar as it is rational, and either representsome conduct as necessary for the relevant agents. 'All men must keep their promises' and 'All rational creatures must keep their promises' would seem to be equally necessary,each within its specified domain. We may also wonder why the concept of morality calls for laws of the latter form. Couldn't there be a distinctivelyhuman morality,in which 'All men must keep theirpromises' would count as a law? Finally, we may wonderwhy such a law could not follow a priori from the concept of a man,just as a rule quantifyingover rationalcreatures might follow from the concepts of reason and rationality. 15. Groundworkvi (389), my translation.For reasonsthat will be explainedbelow, I have broughtthis passage into conformity with Paton's translationof the following passage, in which 'gelten fiir' is translatedas 'hold for'. 16. Groundwork76 (408). THEVOICE OFCONSCIENCE 65 Note, however,that Kant'sexample of absolutenecessity is not a generalrulethatquantifiesoverall rationalcreatures.His example is rathera second-personcommand,'Thoushaltnot lie'. And what Kant says about such a requirementis not that it must refer to all rationalcreaturesbut that it must 'hold for' them-an expression thathe repeatsthroughoutthe Groundwork,as we shall see. Of course, the pronounin 'Thou shaltnot lie' might be standing in for a universalquantifier,andwhat'sat issue could be the domain of that implicit quantifier.Yet if the issue were whether 'thou' referredto all men or to all rationalcreatures,then Kantwouldn't ask for whom the rule holds. The rule, fully spelled out, would be either '(All) thou (men) shall not lie' or '(All) thou (rational creatures)shall not lie', and in eithercase it would have to hold or not hold, withoutlimitation.'All men shallnot lie' cannothold only locally or selectively, any more than 'All men are mortal'. Suppose, however,that 'Thoushaltnot lie' were a type of which various tokens were addressed to various agents, with corresponding variancein the reference of the pronoun.Commandsof this type could be said to 'hold for' particularagentsin two related senses: they might be authoritativefrom the perspectives of particularagents as addressees, and they might consequently be valid in applicationto those agents.To ask for whom the ruleholds would be to ask who finds himself addressedby an authoritative command of this type. According to this interpretation,Kant isn't thinking of moral requirementsas universallyquantifiedrules;he's thinkingof them as personallyaddressedpracticalthoughts,of the form 'Thoushalt not lie'. We can now extend the interpretationso as to explain Kant's chain of inferences. For suppose, next, that when Kant insists on the 'absolute necessity' of moralrequirements,he means thatthe corresponding thoughtmustbe absolutelyauthoritativefromthe perspectiveof the addressee:an agent should not be able to exempthimself from the force of such a thought. Absolute necessity, so understood,can indeed be said to follow from the very concept of a moralrequirement. So we have accountedfor the first link in Kant'schain. Now suppose that 'Thou shalt not lie' would be absolutely authoritative,in the requisitesense, if and only if it were what any agent would think to himself upon consideringwhetherto lie, and 66 J. DAVIDVELLEMAN would think that any agent would think,and so on. If it were such a thought,then an agent consideringwhetherto lie would not only thinkto himself 'Thoushaltnot lie' butwould also thinkof himself as having nothing else to think,because this thoughtwould strike him as what anyone would thinkon the subject,includinghimself on other occasions. He would thereforethink of the question as having been settled once and for all-or, in other words, authoritatively.By contrast,if 'Thou shaltnot lie' weren't such a thought, then even an agent who thoughtit would regardit as optional,there being other things that anyone, includinghimself, might think on the subject.He would thereforefind it lackingin authority.Here is a sense in which the absoluteauthorityentailedin the very concept of moral requirementscan be seen to consist in their 'holding for' all rational agents-that is, by constituting what anyone would think,or would thinkthatanyone would think,and so on. We have now accountedfor the second link in Kant'schain.17 The thirdlink follows withoutfurthersuppositions.The form of what anyone would think, and would think that anyone would think, and so on-the form, if you like, of thatthanwhich thereis nothingelse to think-is the form of a priori knowledge. When it attachesto a thoughtsuch as 'Thoushaltnot lie', it yields a thought that is simultaneously a priori and practical. Hence the very concept of a moral requirementcan be seen to entail an absolute authoritythat is found only in a priori practicalthought.18Kant's argumentis now complete. I have embroidered this interpretive hypothesis on two mere swatches of text. How it will look against the broaderfabric of Kantianethics remainsto be seen. First,however, I wantto register an importantqualification. My hypothesis is that moral laws, for Kant, are not universally quantifiedrules butratherpersonallyaddressedpracticalthoughts, whose universalityand authorityboth consist in their being what anyone would think,andwould thinkthatanyonewould think,and so on. Yet if 'Don't lie' is universalin this sense, then everyone in the relevantcircumstanceswill find himself with nothing else to 17. See also Groundwork92-3 (425): '[D]uty has to be a practical,unconditionednecessity of action; it must thereforehold for all rationalbeings...'. 18. See Groundwork 93 (426): 'These principles must have an origin entirely and completely a priori and must at the same time derive fromthis theirsovereign authority....' THEVOICE OFCONSCIENCE 67 think;and if everyone in the relevantcircumstancesfinds himself with nothing else to thinkbut 'Don't lie', then there will, in effect, be a universalrule of not lying. For this reason, my hypothesis cannot be that moral laws, for Kant, aren'tuniversallyquantifiedrules at all; it must be that they aren'tuniversallyquantifiedrules in the firstinstance.Morallaws, as I understandthem, can be expressed in universallyquantified rules, provided that those rules are understoodas expressing the authority of personal practical thoughts, whose authority just consists in theirbeing whatanyonewould thinkthatanyonewould think. Let me emphasize, then, thatI do not mean to ignore or dismiss the many passages in which Kant himself enunciates laws as universallyquantifiedrules of behaviour.I merely suggest thatthe universal rules enunciated by Kant should be understood as summariesof somethingmorecomplex, or as the outersurfacesof something deeper-namely, a state of affairs in which practical thoughts, in personal form, are common knowledge among all agents. How universalizationworks.Withthis qualificationin mind,I want to apply my interpretive hypothesis to Kant's account of universalization,the procedureby which maxims are tested under the categorical imperative. Here, too, the hypothesis helps to resolve both textualand philosophicalproblems. Considerthis instance of universalization:19 [A person]findshimselfdrivento borrowingmoneybecauseof need.He well knowsthathe will notbe ableto payit backbuthe sees too thathe will get no loanunlesshe givesa firmpromiseto pay it back withina fixed time. He is inclinedto makesuch a promise;buthehasstillenoughconsciencetoask'Isit notunlawful and contraryto duty to get out of difficultiesin this way?' Supposing,however,hedidresolvetodoso,themaximofhisaction wouldrunthus:'Whenever I believemyselfshortof money,I will borrowmoneyandpromiseto packit back,thoughI knowthatthis will never be done.' Now this principle of self-love or personal advantageis perhapsquite compatiblewith my own entire future welfare; only there remains the question 'Is it right?' I therefore transformthe demand of self-love into a universallaw and frame 19. Groundwork90 (422). I have broughtPaton's version of this passage into conformity with his translationof the precedingpassage, by rendering'gelten' as 'to hold'. (See note 15, above.) 68 J. DAVIDVELLEMAN my questionthus: 'How would things standif my maxim became a universallaw?' I then see straightaway thatthis maxim can never hold as a universallaw of natureand be self-consistent, but must necessarilycontradictitself. Forthe universalityof a law thatevery one believinghimself to be in needcan makeanypromisehe pleases with the intentionnotto keep it wouldmakepromising,andthe very purposeof promising,itself impossible,since no one wouldbelieve he was being promisedanything,but would laugh at utterancesof this kind as empty shams. The targetof universalizationin this passage is what Kantcalls a maxim of action: 'WheneverI believe myself short of money, I will borrowmoney andpromiseto pay it back, thoughI know that this will neverbe done.' We might thinkthatthe way to make this maxim universal is to replace the first-person pronoun with quantified variables ranging over all rational creatures.20Kant seems to suggest such a procedure when he refers to 'the universalityof a law thatevery one believing himself to be in need can make any promise he pleases...'. But Kant also suggests a different procedure,when he considers whetherhis maxim itself 'can... hold as a universallaw'. Kant'smaxim is framedin the first person, and so it-the maxim itself-can 'hold' as a universallaw only if first-personalthoughtscan somehow be universal. Kant's framinghis maxim in the first personis no accident.He could not have restatedit, for example, as 'ImmanuelKant will make lying promises when he is in need'. Such a third-personal thoughtwould not be a maximof action,since it could not be acted upon by the thinkeruntil he reformulatedit reflexively,in the first person. Insofar as the target of universalizationis a practical thought,it is essentially first-personal.21 20. For an interpretationof universalizationalong these lines, see e.g. Onora O'Neill, Acting on Principle: an Essay on KantianEthics (New York:ColumbiaUniversity Press, 1975), esp. ChapterFive, 59-93; and 'Consistencyin Action', in Constructionsof Reason; Explorations of Kant's Practical Philosophy (Cambridge:CambridgeUniversity Press, 1989), 81-104. See also Christine Korsgaard,'Kant's Formula of Universal Law', in Creating the Kingdomof Ends (Cambridge:CambridgeUniversity Press, 1996), 77-105. According to Korsgaard,universalization'is carriedout by imagining, in effect, that the action you propose to performin orderto carryout your purposeis the standardprocedure for carryingout thatpurpose'(92). In the presentcase, then, the agent 'imagines a world in which everyone who needs money makes a lying promiseand he imaginesthat,at the same time, he is part of that world, willing his maxim' ('Kant's Analysis of Obligation: the Argumentof GroundworkI', in ibid., 43-76, at 63). Finally, see Roger J. Sullivan, Kant's Moral Theory(Cambridge:CambridgeUniversityPress, 1989), 168-69. 21. On this topic, see JohnPerry,TheProblemof the EssentialIndexicaland OtherEssays (New York:Oxford UniversityPress, 1993). THEVOICE OFCONSCIENCE 69 This first-personalthought should remind us of the secondpersonal injunctionconsidered above, 'Thou shalt not lie', which was there regardedas being addressedby the agent to himself. So regarded,'Thou shaltnot lie' was couched in what might be called the reflexive second-person-the second-person of talking to oneself. And when it is thus addressedto oneself, 'Thou shalt not lie' is just the contradictoryof 'I shall lie', the maxim that is currentlyup for universalization.Ourearlierreflectionson how the second-personal injunction could be a universal law are thus directlyrelevantto the universalizationof the first-personalmaxim. As before, we might consider transformingthe maxim into a universal law, by substitutionof a quantifierfor the first-person pronoun. But Kant speaks more often of maxims' being laws themselves than of their being transformedinto laws. In addition to asking whethera maximcan 'hold as a universallaw',22he asks: whethermaxims can 'serve as universallaws',23 whetherthey have 'universal validity... as laws'24 or 'the universality of a law'25 whether a maxim 'at the same time contains in itself its own universalvalidity for every rationalbeing'26or is constrained'by the condition that it should be universallyvalid as a law for every subject';27whetherit 'can have for its object itself as at the same time a universallaw'28or can 'have as its contentitself considered as a universal law'.29 All of these expressions call for a single thoughtto be regardedsimultaneouslyas the maxim of one agent and as a law for all. According to my interpretation,however, a single thought can simultaneouslybe a first-personalmaxim and a universallaw, if it is what anyone would think in response to the relevantpractical question,and would thinkthatanyonewould think,and so on. It is then a type of thoughtwhose tokenswould be authoritativefor any agent. And imagining that 'I will make false promises' would be 22. 23. 24. 25. 26. 27. 28. 29. Also at 103-4 (438). 94 (426). 126 (458); see also 129 (461). 128 (460). 105 (437-38). 105 (438). 114 (447). 115 (447). 70 VELLEMAN J. DAVID authoritativefor anyone is a way of imagining a universallaw of making false promises. This interpretationexplainshow an individualmaxim can 'have as its content itself consideredas a universallaw'30 or 'containin itself its own universal validity for every rational being'.31 Universalizing a first-personal maxim ('I will make false promises') is not, in the first instance, a process of conjoining it with some universally quantifiedvariantof itself ('Everyone will make false promises'). Universalizing this maxim is rather a matterof regardingthe maxim itself as what anyone would think, or would think that anyone would think, and so on. The universalized maxim is more like this-'Obviously, I will make false promises'-where 'obviously' indicates that the following thoughtwould occur to anyone, as would occur to anyone, and so on. That's how a first-personal maxim can contain its own universal validity within itself. Kantsays thata universallaw of makingfalse promiseswould have the result that 'no one would believe he was being promised anything, but would laugh at utterances of this kind as empty shams'. If we think of this law as a universallyquantifiedrule, to the effect that everyone may or will make false promises when in need, then we shall have to wonderwhy it would have the results predicted. The answermightbe thatpeople's adherenceto sucha law would entail the issuance of so many false promisesthateveryone would eventuallylearnto distrusteveryoneelse.32But this answerwould be a piece of empirical reasoning, about how social interactions wouldevolve in responseto a particularpatternof conduct;whereas Kant says that the requirementsof morality must be derivablea 30. 115 (447). 31. 105 (437-38). 32. For this interpretation,see O'Neill, 'Universal Laws and Ends-in-Themselves', in Constructionsof Reason, 126-44, at 132: 'The project of deceit requires a world with sufficient trustfor deceivers to get othersto believe them;the resultsof universaldeception would be a world in which such trust was lacking, and the deceiver's project was impossible.' See also Korsgaard,'Kant's Formulaof UniversalLaw', 92: 'The efficacy of the false promise as a means of securing the money depends on the fact that not everyone uses promises this way. Promises are efficacious in securing loans only because they are believed, and they are believed only if they are normallytrue.' Finally, see Sullivan, Kant's Moral Theory, 171: 'Truthful assertions cannot survive any universal violation of the essential point of such speech. Once everyone lies for whateach considersa "good"reason, we can never know when any verbalbehaviorcounts as "tellingthe truth".' THEVOICE OFCONSCIENCE 71 priori. This piece of empiricalreasoningwould thereforebe out of place in the process of universalization,by which the specific requirementsof moralityare derived. What's more, the same empiricalreasoningwouldn't apply to a law licensing promises whose falsity would go undetected,since the proliferation of undetectably false promises would not underminepeople's trust; yet Kant reaches the same conclusion about a law of undetectablefalsehoods. He imagines a case in which 'I have in my possession a deposit, the owner of which has died without leaving any record of it'. Moral reflection in these circumstancesraises the question 'whetherI could... make the law that every man is allowed to deny that a deposit has been made when no one can prove the contrary'.Kant's conclusion is 'that taking such a principleas a law would annihilateitself, because its result would be that no one would make a deposit'.33 This conclusion cannot be an empirical prediction of what would happenundera universallyquantifiedrule of denying unrecorded deposits. General adherence to such a rule would not in fact discourage prospectivedepositors,precisely because there would be no recordof the deposits involved. In my view, however, the way to imagine a universal law of denying unrecordeddeposits is to imagine that the maxim 'I will deny unrecordeddeposits' is authoritative,in thatit is whatanyone would think, and would thinkthatanyone would think,and so on. This law would indeed undermine the faith of prospective depositors-not empirically, through the pattern of conduct it produced;butrationally,throughthea priori practicalthinkingthat it embodied, which would be common knowledge among all agents. No one would make unrecordeddeposits if stealing them were all there was to thinkof doing with them. If the maxim of denying unrecordeddeposits were a law in this sense, then the authority of that maxim would be evident to prospectivedepositorsno less thanit was to theirintendedtrustee, since the maxim would be what anyone would think that anyone 33. Critiqueof Practical Reason, trans.by Lewis WhiteBeck (Indianapolis:Bobbs Merrill, 1956), 27 (27). The same case appears,with embellishments,in the essay 'On the Proverb: That may be True in Theory, But Is of No PracticalUse', in Perpetual Peace and Other Essays, trans. Ted Humphrey(Indianapolis:Hackett Publishing, 1983), 61-92, at 69-70 (286-287). 72 J. DAVIDVELLEMAN would think. Depositors would only have to reason aboutthe case from the perspectiveof theirtrusteein orderto see whathis maxim for dealing with their deposits would be, since there would be nothing else to think of doing with them. That the trustee would deny having receivedtheirdepositsisn't somethingthatdepositors would have learned from past experience of his or anyone else's behaviour;it's something that would be evident to them through their own practical reasoning, as proxy for his. They would consequentlybe deterredfrom makingunrecordeddeposits. This interpretation simply assumes that the connections fundamentalto Kant's conception of morality-the connections among universality,necessity, and the a priori-hold for all of the laws involved in universalization,including: (1) the categorical imperative, in which the procedure of universalization is prescribed;(2) the specific requirementsderivedby means of that procedure;and, finally but crucially,(3) the laws imagined within it. In this last instance, imagining one's maxim to be a universal law must entail imagining it to have all three connected properties-that is, to be universallyinescapablea priori. Hence universalization is a procedure of imagining one's maxim to constitutepracticalbut a priori and hence common knowledge. The nature of maxims. Thus far I have avoided inquiringinto the nature of maxims, choosing instead to work with simple expressions of intent, such as 'I'll make false promises' or 'I'll deny unrecordeddeposits'. Now thatI have offered an hypothesis as to how maximsareuniversalized,however,I can no longeravoid the question of what they are and, more importantly,why they mightbe subjectto such a procedure.And I don't thinkthatmaxims are simply intentionsor expressions of intent. Kant says that maxims are 'principles of volition'.34 Many interpretershave noted that Kant usually formulates maxims of action so as to specify both a type of behaviourand a purposeto be servedby it-or, in otherwords,an end as well as a means.35I think thatmaxims so often connect end andmeans, anddo so in the form 34. Groundwork68 (400). 35. See O'Neill, Acting on Principle, 37-38; Korsgaard,'Kant's Analysis of Obligation', 57-58, and The Sources of Normativity(Cambridge:CambridgeUniversity Press, 1996), 108. THE VOICEOF CONSCIENCE 73 of general principles, because they state the connection between reasons and action.36 Consider again the maxim of a lying promise: 'Whenever I believe myself short of money, I will borrowmoney and promise to pay it back, though I know that this will never be done'. I interpretthis maxim to mean that financial need is a reason for promising to return a loan, and that this reason outweighs the countervailingconsiderationthatthe promise would be false. The maxim is thus a principleof volition in the sense that it licenses a practicalinference, from the premises 'I need morney'and 'I'd be lying if I promisedto repaya loan', to the conclusion 'I'll promise to repaya loan'. The license for this inferenceis framedas a general principle because the validity of an inference-type cannot vary from one token to another. More importantly,the validity of an inferenceis a logical relation thatmustbe recognizablea priori. That's why a maximis naturally subject to the test of universalization.If there is a valid inference from 'I need money' to 'I'll makea false promise',thenthe validity of that inference must be such as anyone would recognize, and would recognize that anyone would recognize, and so on. The validity of a practicalinference,like the validityof modusponens, must hold for-and be commonknowledge among-all thinkers. In this case, the inference can't be valid, precisely because its validity would have to be common knowledge, which would underminea presuppositionof the inference itself-namely, that making false promises is a means of getting money.37If it were common knowledge that a decision to make false promises followed from a need for money, then nobody would lend on the basis of promises;promiseswouldn'tbe a meansof gettingmoney; and a decision to make them would no longer follow. Thus, if 'I'll make false promises' did follow from 'I need money', then it 36. See Korsgaard,'An Introductionto the Ethical, Political, and Religious Thought of Kant', in Creating the Kingdomof Ends, 3-42, at 13-14: 'Your maxim must contain your reason for action: it must say what you are going to do, and why'; 'Kant's Analysis of Obligation', 57: 'Your maxim thus expresses whatyou take to be a reasonfor action.' I am inclined to put a slightly finer point on this claim, by saying thatthe maxim states the rule of practicalinference, from reasonto action. 37. Here I follow what Korsgaardcalls 'the practicalcontradictioninterpretation'('Kant's Formulaof UniversalLaw', 92). I differ from Korsgaard,however, in tracingthe practical contradictionto an imaginedpiece of commonknowledge ratherthanan imaginedstandard practice.(See note 20, above.) 74 VELLEMAN J. DAVID wouldn't follow, after all; and so it doesn't follow, to begin with. A desire for money isn't a valid reasonfor makingfalse promises. Its not being a reason is also a priori. And this point provides the most challenging twist in Kant's argument.Kant thoughtthat we cannot wait passively to receive practical dictates with a priori authority,andhence thatwe cannotwait for the voice of conscience to speak.38We have to proposeour own practicaldictates and ask whether they could possibly carrya priori authority.And sometimes, when the answer is no, that answer turns out to carry the sought-forauthority:it resoundswith the voice of conscience. The practicaldictate in the present example is the maxim that making a false promise follows from circumstancesof financial need. That the validity of this inferencemust be a priori is itself a priori, since validityis a matterof rationality,which is common to all thinkers.From the a priori requirementthat the validity of an inference must be a priori, the impossibility of a valid inference from financial need to false promises follows a priori as well. Anyone can see, and can see that anyone can see, that the validity of this inference would have to be a priori, but that one of the inference's presuppositions would then be false, so that the inference wouldn't be valid, after all. The fact that the validity of such an inference would have to be common knowledge, which would invalidate the inference-this fact is itself common knowledge among all who care to reflect on the matter.So when the question is whethera need for money is a reason for making false promises, anyone can see that the answer is no, and that anyone can see it, and so on. Here, finally, is a dictate of conscience, reverberatingwith the appropriateauthority.Conscience tells us that the reasons we thoughtwe had for doing somethingcouldn't be reasonsfor doing it; and it tells us authoritatively,once and for all. They couldn't be reasonsfor doing it, conscience tells us, becausetheirbeing reasons couldn't be seen, and be seen to be seen, by all. And what conscience here points out to us is somethingthatcan be seen, and seen to be seen, by all. Thus,conscience authoritativelyrevealsthat our proposed reasons for acting couldn't be authoritativeand consequentlycouldn't be reasons. 38. See again the passage quoted in note 14, above. THEVOICE OFCONSCIENCE 75 The role of autonomy.But isn't conscience supposed to forbid us from doing things ratherthanmerely informus thatwe don't have reason for doing them? Kant's answer, I think, would be that by informing us of the absence of reasons for doing things, conscience rules out the possibility of ourdoing themfor reasonsand,with it, the possibility of ourdoing them autonomously-or, indeed,the possibility of our doing them, since we are truly the agents of the things we do only when we do them for reasons.And rulingout the possibility of our being the agents of the things we do is the way that conscience forbids us from doing them at all. Kant says:39 [M]orality lies in the relation of actions to the autonomy of the will.... An actionwhich is compatiblewith the autonomyof the will is permitted;one whichdoesnotharmonizewithit isforbidden. Kant could have put his point differently. An action that is incompatible with the autonomy of the will isn't, properly speaking, an action at all: it's a piece of behaviourunattributable to an agent, a bodily movementin which thereis nobodyhome. So put, of course, the point seems to be thatwe won't do the forbidden thing-or, at least, thatwe won't do it. Yet this point is compatible with the recognition that we might still do the forbiddenthing in the weaker sense of 'do' that includes nonautonomousbehaviour. As I interpretKant, the recognition that we could do something only nonautonomouslydeters us from doing the thing even in this weaker sense. The deterrentforce of this recognitionderives from our reverencefor the idea of ourselves as rationaland autonomous beings. Kant speaks of a 'paradox' with the following content: 'that without any furtherend or advantageto be attained[,] the mere dignity of humanity, that is, of rational nature in man-and consequently that reverence for a mere idea-should function as an inflexible precept for the will.'40 In other words, the prescriptive force of moral dictates is a force registered in our reverence for the idea of ourselves as rational and autonomous beings. Conscience tells us thatif we do something, we shall have 39. Groundwork107 (439). 40. 106 (439). 76 J. DAVID VELLEMAN to do it nonautonomously,withoutreason;andconscience thereby appeals to our reverence for this self-ideal as a motive against doing the thing at all. The Kantian ego ideal. I have now returnedto the idea that Kant resemblesFreudin positing an ego ideal. This ideal is necessaryto motivateouradherenceto the conclusionsthatresultfromapplying the categorical imperative-the conclusions that I have identified with the dictates of conscience. These conclusions authoritatively refuteourproposedreasonsfor acting;butin orderto deterus from acting, they must engage our respect for the conception of ourselves as acting only for reasons. Moral requirementsthus motivate us via an ideal image of our obeying them. I believe that the ego ideal plays a similar role in Freudian theory.41Freudsometimes speaksas if the commandsof the superego are backedby threatsand obeyed by the ego solely out of fear. In fact, however,his descriptionsof the relationsbetween ego and super-ego depend heavily on the ego's admirationfor the superego, as an internalizedobjectof love. And it is in this lattercapacity that the super-egois describedby Freudas being, or as including, an ego ideal. I believe that Freud's theory of the ego ideal can help us to humanize Kant's ideal of ourselves as rationallyautonomous.It can help us to see that what Kant called 'reverence for a mere idea'-reverence, thatis, for 'the mere dignity of humanity'42-is in fact our response to something that we have internalizedfrom real people in the course of our moral development. More specifically, I believe thatthe object of this reverence,the ideal of ourselves as rationallyautonomous,is an ideal that we acquirein the course of loving ourparents,in the mannerdescribedby Freud. But my reasons for this belief will have to wait for another occasion. Departmentof Philosophy Universityof Michigan Ann Arbor,Michigan 48109, USA [email protected] 41. The claims made in this paragraphare defended in 'The Direct Heir of the Oedipus Complex' (MS). 42. Quoted at note 40, above.