'Ought' Does Not Imply 'Can' more

North American Philosophical Publications Ought Does Not Imply Can Author(s): Paul Saka Source: American Philosophical Quarterly, Vol. 37, No. 2 (Apr., 2000), pp. 93-105 Published by: University of Illinois Press on behalf of North American Philosophical Publications Stable URL: http://www.jstor.org/stable/20009989 . Accessed: 27/02/2011 15:09 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=illinois. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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 support@jstor.org. University of Illinois Press and North American Philosophical Publications are collaborating with JSTOR to digitize, preserve and extend access to American Philosophical Quarterly. http://www.jstor.org American Philosophical Quarterly Volume 37, Number 2, April 2000 OUGHT DOES NOT Paul Saka IMPLY CAN A he slogan iswidely held that ought im? to perform you are obliged plies can?that an action only if you have the capacity to un? it. Yet this position, perform suitably derstood, conjoined leads with to contradiction a rather when it is thesis show innocuous To Implication as analytic entailment: In vir? tue of the concepts if you ought involved, to do A then you are capable of doing A. The first explicit statement of this is usu? ally attributed toKant (1797, though it can be found much with earlier, at least implicitly), criticisms and defenses subsequent a in Gowans collected It remains (1987). view: within just the past few years acceptable and libertarians. compatibilists this, I shall focus on two modal that share According gument, contradicted the same to hard determinists and to most underlying to the first, the Epistemic the ought-can principle the doctrine arguments structure. Ar? is popular it has been defended by Sapontzis (1992), Gensler (1996), and Zimmerman (1996); and classical analytic three-valued it is virtually deontic built into the axioms of logic.1 presupposition: as (b) Implication This is simply a by ism; but determinism may, for all we now be true; and therefore we don't know know, can. According that ought to the implies second, ought-can by the of determinism; doctrine said doctrine a synthetic makes it claim; and therefore isn't analytic that ought implies can. These two arguments sections occupy section III, I discuss motivations I and II; in behind the the Analyticity Argument, is contradicted principle the of determin? or supervalued to counterpart classical In cases where you are entailment. not able to do A, implication-as-entailment treats "you ought to A" as false, while im plication-as-presupposition or otherwise meaningless value. This latter approach treats without is advocated it as truth by Hare (1963), by Collingridge (1977), and White (1981).2 by Isaiah Berlin as cited in as synthetic (c) Implication this position because log principle, advocated, posteriori, connection: although (d) Implication I have a priori: never as I cata? in it a seen it is tenable synthetic an addi? I introduce ought-can principle, on the tional argument it based against imperatives, on the epistemological significance central arguments. To be clear semantics of and I comment of my about just what I am arguing I will need to distinguish for, among vari? ous construals of "ought implies can." (a) and causal e.g., as an empirical too may have no This account to notice What's proponents. important later on regarding both this construal and 93 94 / AMERICAN PHILOSOPHICAL QUARTERLY the previous justifiably is that if anyone and correctly maintain one were to and correctly recognize "ought implies can" as construed by (a/b/ If we do this, then it's epistemically c/d). justifiably necessary formally: Premise D(X "We know Now (1): ought to A - X can A) for us; we know it.4 To put it that we implies they would as that ought implies can. (e) Implication CONVERSATIONAL IMPLICATURE3: tO Applied can" the implicature theory "ought implies claims that although the bare fact of your can" even "ought as a synthetic fact, then hold know or at least justifiably being obliged to do A does not imply any comment that to do A, someone's capacity are obliged does have this implication, you at least in typical contexts. No one would bother were evant to refer relevant, unless you could to your obligation and it's generally unless not it rel? that ifX ought toA then X can A." of determinism, regardless to re? it's true or false, appears an unsettled main issue. At any rate there is no evident logical or factual contradic? of whether tion, so far as we now various versions asserting current are, they given epistemically possible. regarding ignorance can be represented minism Premise (2): 0(-iXdoesA-iXcanA) the thesis it. This implicature in Sinnott-Armstrong Thesis about do something account is advocated (1984, 1988a), in our know, of determinism; Forrester (1989), and Pigden (1990). (e), that ought merely implicates a denial of the principle is really un? can" as it's normally "ought implies are At any rate, my criticisms derstood. can, concerned with the stronger versions of the Argu? ought-can principle. My Analyticity at the principle when ment is aimed as analytically true (a, b) or as a construed (a-c), while my Epistemic Argument priori is aimed ported as pur? at (a-d) when construed I am of knowledge. Thus, objects the? concerned with the controversial knowledge, our In particular, deter? necessitarian as follows: "For all we know, cannot do A," i.e., if X does not do A then X determinism has thus far not been definitively Premise leads By refuted. to (3), which (2) is equivalent to (4), from which we get (5): contraposition X does A) if X can do A then X does (3): chiefly ses (a) and (b), I am incidentally concerned with the straw theses (c) and (d), and I am not at all concerned with thesis (e). I. The My Epistemic Argument (2), modal 0(XcanA- "For all we know, do A." By one first proof comes in two versions, and the other to full knowledge referring to mere consid? (all things justification Although distinct, technically and I shall essence, tween them ered). these versions are identical switch are in be? By they sometimes (1), (3), agglomeration (4): and (X - X can A) 0[(X ought to A - X does can A A)] (4), modal 0(X transitivity of - (5): ought to A - X does A) if X ought to A then X for the sake of exposition. by re proceeds the supposition The Epistemic Argument under ductio ad absurdum "For all we know, does A." OUGHTDOES NOT IMPLY CAN I 95 Justification modal modal for the rules used here of lutely that P was true, yet we would be open to the possibility that P was false. Although states may such joint epistemic conceiv? are clearly, when held ably obtain, they under reflection, irrational. If we are rea? thinkers, sonably competent that P is true precisely other all we then we know contraposition, transitivity will and agglomeration, be discussed shortly, at the end of the proof. So far we've been speaking level, using schema a particu? A (an action). Now let's consider case. For X, substitute lar "the president," and for A substitute tell the truth." "always Since at a generic letters X (an agent) and not there is a.prima facie obligation to lie that the president vio? presumably we can say (6). lates at least on occasion, Premise (6) ought to A & -?X does A) always know, P is false.5 = -?0 to fail as a universal (DP ->P) were law of epistemic it holds as an systems, to ideal that all epistemic systems ought and moreover it holds when the to us right now. is licensed by agglomeration, (4) than when, for In short, even if strive for, relativized Line D(X proposition that (DP & OQ) we know hold. This that P holds, - 0(P&Q)?if for all we "We know that the president ought to tell the truth and yet does not." This is equivalent hence to (9). By to (7), hence and Q holds know, then for all we know P and Q jointly is supported by the correspond? in such weak modal ing tautology logics as T (assuming that alethic and epistemic modal similar structure except where logics possess there is special reason to think otherwise); and it seems intuitively undeniable, keeping inmind my foregoing remarks on ideality and on relativization to modestly competent reasoners. to (8), and (6), modal DeMorgan's D-? (^X ought law (7): A) to AvXdoes of - (8): By (7), modal D-? definition (X ought to A - X does A) By (8), duality of D/0 (9): ->0(X ought to A - X does A) X As for lines established might make if epistemic (3, 5, 7, 8), these can all be a variety of ways. First, we in the following claim: plausible at moment t S knows system "It's false that for all we know, whenever ought to A, X in fact does A." But ductio (9) contradicts I reject (1). reassess each (5). Therefore, by re at t that P im? that P, and if S also knows our practice then it follows of plies Q, by at t that attitude ascriptions that S knows for the attitude of open similarly In other words, the following mindedness). to be valid. rules of inference appear Q (and Let's examining carefully (1, 2, 6) and the inferences ing the premises To begin with the latter, line (3-5, 7-9). ~i<> ->P, which ing reductio. is supported If it weren't step of the argument, our reason for accept? (Rl) D(P-Q) DP DQ Given these rules of (R2) D(P-Q) OP OQ inference, then, we classical (9) is established by the equivalence DP = by the follow? true, then for some proposition we would have DP and P, 0 -<P (or else the equally unacceptable ~^DP & ^0 ->P). That is, we would know abso can justify recognizing lines (3, 5, 7, 8) simply by the corresponding 96 / AMERICAN PHILOSOPHICAL QUARTERLY tautologies. ognition (P&-Q) we have (-PvQ)]. of example, the classical (-PvQ) For from tautology our rec? the president has or that we can't ought chosen not some kind know of right to lie, that the president was lie. However, my example for convenience, and other merely serve just as well. So long to know at that someone transgress, then the basic D[(P&-?Q) would examples as we happen some time does structure ply. For This, combined with D(P&^Q) from line (6), allows rule (Rl) to return D^>PvQ), i.e., line (7). our might interpret in terms of objective jus? If DP is than knowledge. for accept? justification are not aware of the we Alternatively, modal operators tification rather read as "we have ing P if we Ripper of my argument would still ap? "We know that Jack the instance, and yet ought not to have killed, did kill" would contradict premises (1) and to handle reformulated (2) if these were tense. Nor can premise on be denied (6) we really aren't respon? the grounds that sible move tarians able. for anything isn't even we do or fail to do. This in the running for liber? and compatibilists, of course; and one it's not very if ought thing, reason? were (even and even if we don't believe justification P)", and if OP is read as "we don't have for refusing P", then the fol? justification seem to rules of inference would lowing be valid. for hard determinists For weakened (R3) P-Q DP DQ These rules clearly hold Now system. cognitive can tell that the perfect (R4) P-Q OP OQ for the omniscient in cases where we thinker then too. is allowed surely we only if X of ought would would on so that "X ought to A" is true does A, then the sense actually be so flimsy that there left for insisting be no motivation the principle under dispute, (1). For another certain than thing, (6) is far more of great contro? (1). Thesis (1) is a matter to make these inferences, are allowed to make them if we know anything at all versy; whereas we surely know in moral that discourse, sadistic murderers ought not to do what fail is not that some they do, or more generally to do what to do. they ought to premise (2), which According to (3), we do logically equivalent know false. the which with The of which of these strategies Regardless we opt for, we arrive at the following prin? if you bear a certain ciple: epistemic a proposition (either D or open-mindedness 0), then knowledge can have justification for bearing that you same attitude toward any equivalent propo? attitude toward sition, whose to us or at least equivalence (cf. Hintikka and (1), toward any proposition is trivially discernible that 1962). I conclude is that determinism certainty necessitarian version of de? entertained chain to be here of derives reasoning thesis, and (12, from in (11, 14) terminism following (10) is a substantive analytic, the logic of (3-5) and (7-9) impeccable, substantive for rejection. that we (2), and are is formally left with the 13) appear are logical consequences. a cause, in where to (6) as candidates event mundane has (lO)Every a mundane event stands contrast Premise grounds (6) might be denied never on the that the president lies, or that events such as the cosmic initializing atheist's uncaused origin of the universe or the theist's uncaused origin of God. OUGHTDOES NOT IMPLYCAN I 97 to the mundane, ourselves Restricting P happens only if a sufficient cause then, of P happens: P - cause(P). ( 11 ) If a sufficient cause of an event happens then it is nomologically that necessary that event happen: cause(P) - DP. (12)Combining p _> O^P remarkable from indeed if it could be settled the armchair. II. The Analyticity Argument So far I have (10, DP, 11) gives P contraposition "P" may stand i.e., gives for -,<o -,p whence -?P. Since operators: know" (or alternatively, "all things consid? we have justification for asserting ered, we lack that" and "all things considered, for denying that"). But it is justification to take the modal also appropriate opera? tors as expressing and analytic necessity truth ("it is a conceptual analytic possibility that" and "it is not a conceptual falsehood In this case we get precisely the that"). same form for. of argument supporting as in (1-9), though is called different commentary taken D and 0 as epistemic "we know that" and "for all we "^Ax", (13)If it follows that OAx - Ax. x has the power to do A then it's that x does A (if x nomically possible can do A then it can be the case that x does A): x can A - OAx. together (12, - x 13): x can A (14)Putting does A. Premise (10) is held by incompatibilists and by generally Other compatibilists compatibilists. will demur, but even then most should agree that (10), regard? less of whether it be true or false, is not some current knowl? false given demonstrably one may In other words, to refuse edge. The proof once again proceeds by reduc tio ad absurdum, but this time under the can under that ought supposition implies or (b)?implication as either construal (a) or as analytic entailment presup? analytic position. posited Premise If the ought-can as an analytic truth, (L) ought to A - X can A) that if X ought of an principle then: is believe there (10) and hence (14); but, since is no proof once acceptable modal it, (14) becomes against it is subsumed under the we know." Accordingly, "for all line D(X is established. (3) in the main argument to turn my reduc Some will be tempted on (1), they will tio around; insisting conclude that (2, 3) are false and that de? terminism result, be highly it would Epistemic is deductively if anyone were significant preserve This disprovable. to defend it, would in its own right, and the interest of the I do not find it at all "It is analytically necessary to A then X can A." Now the thesis of determinism, regardless whether it's true or false, appears tomake con? is analytically claim, hence empirical is analytically and hence tingent, possible. In particular, the claim of necessitarian de? terminism Premise can be represented (2') 0(-iKdoesA^XcanA) as follows: Argument. however. it won't First, ap? convincing, to those who regard (10) as possible peal at least in principle. Second, as a claim of knowledge, less se? (1) is prima facie a claim cure than (2/3), of ignorance. the question of determinism Third, ap? pears to be a scientific one. It would be "It is analytically possible that if X does not do A then X cannot do A," i.e., "if X does not do A then X cannot do A" is not false by-definition. 98 / AMERICAN PHILOSOPHICAL QUARTERLY Premise leads By (2') to (3'), which is equivalent from which we get (5'): contraposition X does A) (4'): can A) and (3'): x is a murderer murder...). then x ought to refrain so, though it would Doing yields consider where effectively from com? to (4'), (2'), modal 0(XcanA- plicate matters, end results. To vised see this, the same re? are the following lines By (L), (3'), agglomeration to A - X 0[(X ought - X does (X can A A)] By (4'), modal 0(X Now we transitivity (l"-8") proof, on grounds to those analagous given for (1-8) and (l'-8'). - x can (1") DVx(x ought to A A) licensed (2") (3") (4") OVxi-'x OVx(x does A ->x can A) of - (5'): ought to A - X does A) a counterpart to (6) I propose that for can A - x does A) need to find analytically. and for A we X, we substitute "murderers," substitute "refrain from murder." It is an analytic frain from truth that murderers ought for murder to re? is by that holds - x can OVx[(x ought to A A) and (x - x does can A A)] OVx(x ought DVx(x DVx-^-'x DVx-i(x D3x->(x ought to A - x does A) to A and -'x does A) to A v x does A) to A - x does A) to A - x does A) x does A) to A - x does A) (5") (6") (7") (8") (8.1) murder, definition a kind of wrongful killing. It is also an analytic truth that murderers do not refrain from murder, for a murderer by is one who murders. definition Premise (6'): ought to A and ->X do A) true that murderers and yet do not." ought ought ought (8.2) (9") D^Vx(xoughttoA^OVx(x move from ought D(X "It is analytically to refrain from ought murder The critical inference cannot tell Granted, analysis conceptual whether is in fact any particular killing or whether in fact com? anyone wrongful, mits murder, and so it is not analytic whether whether murder and murderers exist. But or not murderers exist, that murderers necessary ought and yet do so, just as it is an it remains (8") in this proof lies in the to (9"). (9") is justi? fied given the definitional duality of D/0 and (8.2); (8.2) is justified given the defi? nitional duality of V/3 and (8.1); and (8.1) can be justified (8") so long as it's given true that the universe is non? analytically Now in standard model set theory, in standard again is non-empty. it is a Hence, canonically, an analytic and therefore truth, logical is non-empty. truth, that the universe to treat it is perfectly standard Although as an of the universe the non-emptiness feel that this truth, you may analytic serves more to undermine standard premise logic than to support (l"-9"). the revised ity Argument Analytic I sympathize, and theory, and the universe empty. analytically not murder their nonexistence, truth, despite analytic are simians that do that winged monkeys If you remain unmoved have wings. by this and persist consideration, istential presupposition would in seeing an ex? in (6')?which invalidate its claim to analyticity? then you are free to replace my schematic structure X with a quantificational ([Vx] if OUGHTDOES NOT IMPLYCAN I 99 two other ways like to suggest we it. In the first place, of defending the proof conditionalize somehow might so as to conclude is that if the universe cannot analyti? ought can. In the second place, cally imply or not the non-emptiness of the whether non-empty, then is a logical truth, it is surely an reason a priori truth (for the Cartesian ana I think, I exist). Whether that, since a priority, of follows upon lyticity universe course, versy. Analyticity otherwise, rectly is a matter If it does, of enormous then the contro? revised But so Iwould Now hence By given (8'), premise and hence (6'), we (9'). law (7'): have (7'), (6'), modal DeMorgan's D -i( -?X ought to A v X does A) of - (8'): By (7'), modal D^(X definition ought to A - X does A) (9'): By (8'), duality of D/0 ->0(X ought to A - X does A) can stand as it is; Argument di? the Analyticity Argument an A Priori converts into as (9') contradicts (5'). Therefore, I reject (F). in the case of (1), by reductio The Analyticity and Epistemic Argu? ments the same formal precisely structure. in their however, differ, They In and in their vulnerabilities. conclusions share the first serves we the Epistemic Argument place, as a reductio against the premise that for either know or have justification and thus works can," implies it is the ought-can when principle is not that determinism given Argument: a priori im? it follows that "ought false, true. This A is not a priori can" plies Priori Argument is Analyticity goes only also Argument not the ought-can principle against and (b), but under construals (a) it is weaker (c). At the same time, Argument ought-can determinism in its ten proponents is false a than the stronger for it in its scope, "ought against than the Analyticity for some ability, may insist that under any of the construals (a recognized In contrast, the Analyticity d). Argument works (c). against only (a, b), and maybe On ment hand, the Analyticity Argu? is less open to the charges brought the then countered) up (and against These Argument. charges?in? Epistemic the validity of the inference steps and the tenability, for the hard even of premise determinist, (6)?wouldn't volving (3-5, 7-9) arise ment. Argu? is the Analyticity Argument Finally, immune to empirical that might discoveries as urged by lead us to reject determinism, some interpreters of quantum For physics. if we knew determinism were a posteriori if our knowledge ought-can construed to be false, then the in the case of the Analyticity the other this is not the usual posi? Since priori. the tion of the anti-determinist, however, as A Priori Argument remains practically as the Analyticity interesting I have offered In summary, possible which denies defenses of premise that existence Argument. a number of (6'). The first, is necessarily presupposed by "murderers ought to refrain is one that I personally fa? from murder," a vor; the second, whereby logic posits follows from standard universe, non-empty model the third, theory; regarding is intuitively conditionalization, moving to develop difficult for? (though, I confess, and the fourth the preserves mally); soundness of the "Analyticity" Argument it as the slightly weaker A by interpreting Priori Argument. even would still fail when principle a priori as an truth; and simi? were synthetic if our knowledge then larly, the ought-can would still fail principle construed as an analytic truth. when 100 / AMERICAN PHILOSOPHICAL QUARTERLY III. The Reductions General We Argument Comments and also serves an evaluative function whereby to some? it ought judgment to guide one adopts an emotional relation one or something ("reality bites; not be that does the way not necessarily it is" is a moral function are consequently left with the rejec? tion of two related theses: (1) that we are justified can; and analytic. contributes in the belief that ought implies (F) This that "ought can" is implies What raises the question, to to the widespread intuition In the second place, it is a mis? behavior). take to suppose that function is overriding, that structure is irrelevant. Ought very pos? what sibly means to "the" reference ought-stalements. it means purpose of without issuing the contrary? be For one, "ought can" might implies a legitimate moral the object of imperative or report. That is to say, though "ought can" is not true, it arguably implies to be (cf. Moritz Brown 1968, Marcus 1981, Kekes 1980, White Rescher 1987, and Forrester 1989). other words: (15) In the perfect world, X can A). strikes fails me (X ought to A -> ought 1977, 1984, In yet defenders of ought-can, I realize, Many will remain unswayed such general con? by siderations 1797, Sapontzis (e.g., Kant insist that the ought-can 1992). They will principle between But how stood? by some connection reason. and practical obligation to be under? is this connection is validated In what follows, reductive I shall argue that This ought mative tangle possible of their as correct. can within to imply (for instance, systems that make of inconsistencies for some Although nor? many the law is a it im? analyses generally "ought implies can." falsify view grounds To begin with, the threat to avoid harm in the motivation obligation to oneself. This view Kanger-Anderson logic to ordinary to do A in the is expressed of deontic reduction the available systems often implicitly of what taken to discharge all agents I would such regard legal duties), as unjust. Since ethical theorizing to be in the context takes place the case, it is easy to ought see how someone might the mis? expound thesis modal you logic, whereby iff it is necessarily the case ought that, if you don't do A, you will suffer some (Anderson 1958). If penalty or misfortune then of ought, however, this is the analysis must for it is the ought-can fail, principle is all too often perfectly possible?and one the case?for actually less of what one does. The practical connection reason between might to suffer regard? of the (1) under the influence thesis (15). It is furthermore easy plausible to see how someone might unconsciously move from (15)to(l) of the panglossianism civilization.6 ought-can if we attractive The under the influence that pervades our and obligation a instead reflect principle suppose and is to guide behavior judgment to motivate of reward and punish? systems can be ment. This however, supposition, a it is always In the first place, criticized. of moral mistake to suppose has a that anything as a guide function. Besides serving unique to behavior, for example, moral judgment also seem might that the function between ought state? meaning-equivalence ments and imperatives, (16) is whereby to (17) (cf. Kant 1797, Hare equivalent 1952, and Gensler 1996). (16) (17) But You ought to take off your shoes. Take off your shoes! im? too contradicts "ought can." First, the simple transformational plies this reduction OUGHTDOES NOT IMPLY CAN I 101 analysis treats (17) as equivalent et al. 1973). But is equivalent to (18) if this is instance of the conceivable truth particular that you ought to save the lives of all hu? manitarians. For these the reasons, of pointlessness mands suggests treats impossible-to-fulfill at that ought de? best (e.g., true, then (16) is absurd. (18) You will Stockwell to (18), which take off your shoes. analysis the performative Second, as equivalent to (19) (17) conversationally implicates can (cf. Sinnott 1984 and 1988a and Forrester Armstrong 1989; though see Rescher 1987, p. 40). To summarize, the factual ought-can be held either as a careless principle might of the ideal ought-can paraphrase principle or as a consequence of reductively (15) reason? in practical grounding obligation for ing. But rehearsed?the tive cum the reductive theory, accounts the impera? the analysis, threat (e.g., McCawley 1968, Lewis 1972), in which case (16) is equivalent to (19). But if this is true then ought order imply can, for I can easily to do the impossible, futile you though that be. does not (19) (20) I order you to take off your shoes. I order this: you take off your shoes. treats (17) of sentences tranformational analysis the paratactic Third, as equivalent to the sequence in (20) (e.g., Davidson true then once can, Under 1979). But if this is cum performative and imperative analysis, cum paratactic the imperative analysis? none is consistent with "ought implies can." serve These considerations, therefore, as further justification, I dub which the Reductions for rejecting the Argument, again ought does not imply and for the same reason as before. the latter and the analyses, the paratactic, it is ad? performative true that there is often no point to mittedly un that you ought to do something saying you observation principle can be good less are for able two to do reasons. so. For But this fails to secure the ought-can one, there ought-can principle. To return tomy earlier, modal arguments, a common I find, is "Of course response, the thesis that ought implies can is incom? with determinism. Determinists patible don't statements. Your recognize ought proof is trivial." But this view is mistaken in two ways. In the first place, ism is compatible with moral Because ence of hard determinists free will, it determin? realism. the exist? reason to demand the impos? sible from you, as when such demands will spur you on to superior performance. Therefore the pointlessness is a charge half-truth. ness For another charge thing, the pointless is irrelevant, for the an utterance a context in of it false. For inmost instance, for me to mention of China. inability (because Like? to save she is pointlessness does not make contexts they no means inevita? by sometimes?though that agents are accountable for bly?deny their errors. But even when determinism is supplemented still falls short sors ought such ought important determining in this way, the position of denying that transgres? not do what they do. For to deny statements force?social factor is to withdraw pressure?as in others' behavior. an one deny is true, it is pointless that Beijing is the capital of China; yet is in fact the capital Beijing in the context of your wise, the life of Margaret S?nger mention theless already dead), it is typically pointless to to do so. None? your obligation true that you ought it is conceivably to save the life of S?nger, as that may be a that hard determinists have so? Assuming cial wants, it would their go against interests to deny ought statements. For this 102 / AMERICAN PHILOSOPHICAL QUARTERLY reason, I don't think that hard determinists (24) If X ought to A then X can A, but we it. don't justifiably believe and in point should deny ought statements; do. of fact, I don't think that they generally Far more the triviality criti? importantly, cism misses not assume I do the point of my proofs. the truth of determinism, only of deter? The mere possibility to overthrow While this position is possible for the un? no it deserves reasonable dogmatist, its cogency. suffices minism attention. Thus any success? philosophical ful proof (22) leaves against legitimate to as? of the dialectic unable participants (21). Given assert (21), who In order result. consider (25) (26) sert the prevalence of those this is a highly significant to grasp this point better, to religious thought. the necessity can." Thus, even liber? of "ought implies that tarians can accept my proofs?proofs to the growing that literature contribute to the ought-can takes exception principle van Fraassen 1973, 1987, 1984, Rescher and 1984 an analogy God We exists. know 1962, (cf. Lemmon Marcus Kekes 1980, that God it). exists (we justifi? 1988, Sinnott-Armstrong Forrester 1989, and Horty 1994). Another criticism of my position points out that the Epistemic is hardly Argument not (21) because it contradicts interesting ably believe but (22). (21) (22) If X ought to A then X can A. there is a significant difference Of course between the case refuting (25), i.e., proving and refuting for strong atheism, (26), i.e., the case for non-theism (which proving encompasses cism). But and agnosti? strong atheism even so, any refutation of (26) where in a society significant We now know that ifX ought toA then for X can A, or We have justification that if X ought to A then X believing can A. is that the Epistemic Argument in this way, and it's also true that For even if the this limitation is significant. It's limited Epistemic Argument It is false can A. true is highly claims for (25) are popular. The interest of my work generalizes be? for both of my arguments ought-can, yond are essentially of (R5), a valid instances schema case, (In my having wide application. = P = the thesis of determinism, Q principle.) (R6) O(P---Q) is sound, it does noth? the ought-can (R5) P--Q ing to establish (23): (23) that if X ought to A then X OP P -OQ My sister claim -OQ is con? So far as the Epistemic Argument of "ought cerned, therefore, my refutation a I would can" call is what implies counter-proof rather than a disproof. That said, to deny the essential interest for of the that the schematic (R5) and its are both valid is neither (R6) surpris? ing nor new. I see it as my contribution, to emphasize what seems to have however, that (R5) and (R6) can be unnoticed, to work in philosophical put disputations. of ought-can, for Aside from the problem gone consider example, from Evil: the atheistic Argument Argument Epistemic it is pragmatically is to miss the fact that rational of (21) to hold (22) as well, at proponents For if ought-can least implicitly. propo? nents did not hold (22), itwould be possible for them to say: OUGHTDOES NOT IMPLYCAN I 103 (27) If evil exists then a supremely God doesn't. pow? likewise for the mere epistemic possibil? erful benevolent (28) (29) While Evil exists. Therefore some ity of (28), given (27)]. For example, ifwe accept (27), that evil is inconsistent with God's refute and if we existence, that evil exists, (28), cannot hold the mere fail to outright then according that God exists. pos? God does not exist. attempt to prove that rest content (27) (28) is false, others with merely that, for all we know, showing or (28) can be false despite (27) appear? ances to the contrary (e.g., Plantinga theists or to shift the is supposed 1974). This move burden of proof back to the atheist. But, if have been cor? my preceding arguments of (27), possibility epistemic would that we have no (28), prove given or even jus? of God's existence knowledge rect, the mere to (R5) we Analogously, other, would metaphysical sibility of either (27) or (28), given the is not a nec? prove that God are to if theists Therefore essary being. answer the Argument from Evil, they must do more than deny the necessity of pre? (27, falsehood ture of paper tigation Brooklyn 28); they must argue for the of at least one.7The basic struc? arguments contribute subject CUNY used in this to the inves? mises the modal of diverse College, can therefore tification at all for believing inGod [and areas.8 NOTES 1. In "standard" deontic logics, logical can: Since "-0(P&~P)" logically Q ought lated given such as the system of von Wright 1963, ought effectively implies is a theorem, if O applies to some proposition Q then Q can't be this principle is weaker than "ought implies practical can99?"If impossible. Although to be the case, then Q is practically possible"?it is as strong a claim as can be formu? von Wright's representational resources. For a similar theorem, see Aqvist 1984; for deontic logics that take exception, argument against standard deontic see van Fraassen logic, see Forrester 1973, Marcus 1995. 1980, Horty 1994; for an 2. Actually, Hare writes that the operative is merely implication He does not indicate just how the two differ, however. (?4.3). 3. An explanation "analogous" to presupposition of implicature: Suppose that you say "P or Q," for example. Then you conver? sationally implicate that you do not know either P or Q alone. The reason is that you presumably converse according to the Cooperative Principle, with its maxims of relevance, brevity, informa and that such cooperation precludes the statement "P or Q" whenever tiveness, etc., you could as well say "P" or "Q" (Grice 1989). In general, your statement A conversationally just impli? cates B iff your saying 4. Epistemic modality knows at one moment, interpreted relative that A, combined is always another now. with the cooperative maxims, implies B. relative to epistemic systems at specific times: what one subject need not. In the proof that follows, modal operators are to be to us = of paraconsistent -"0-p) logic (e.g., Priest 1987) may deny that (DP always holds (in order to treat the liar paradox for instance). But even paraconsistent logicians = in typical non-paradoxical may easily accept (DP ^O^P), for rational epistemic systems, cases (as found here). 5. Some defenders 104 / AMERICAN PHILOSOPHICAL QUARTERLY live in the best of all possible worlds is arguably a direct corollary of Judeo-Christo Islam and its omnipotent, omnibenevolent God. Contrast this, for instance, with the Buddhist is suffering. In addition there are colonial, technological, doctrine that existence and educational sense of optimism. factors that ground the modern Westerner's 6. That we the impossibility 7. Establishing of (28) would seem to be a daunting task, given that the only obvious way of doing so would be to deduce the denial of (28) from the goodness of an omnipo? tent God, which would be question begging. 8. I am pleased to acknowledge reviewer for helpful comments Walter Sinnott-Armstrong, William on an earlier draft of this paper. Lycan, and an anonymous LIST OF REFERENCES Anderson, Alan Ross. 1958. "A Reduction of Deontic Logic toAlethic Modal Logic." Mind, vol. 67. In Handbook 1984. "Deontic Logic." of Philosophical Aqvist, Lennart. Logic, vol. 2, ed. Dov to Deontic Logic and and F. Guenthner. Dordrecht: Reidel. Reprinted as Introduction Gabbay the Theory of Normative 1987. Systems. Naples: Bibliopolis, Brown, James. 1977. "Moral Theory and the Ought-Can Principle." Mind, vol. 86, pp. 206-23. " D. G. 1977. and Hume's Rule." Philosophy, vol. 52, pp. 'Ought-Implies-Can' Collingridge, 348-51. Davidson, in Inquiries into Truth and Inter? Donald. 1979. Moods and Performances. Reprinted Press, 1984. pretation. New York: Oxford University James. 1989. Why You Should: the Pragmatics Forrester, of Deontic Speech. Hanover, N.H.: Press of New England. University _. 1995. "Conflicts of Obligation." Araenctm Philosophical Quarterly, vol. 32, pp. 31^4. Harry. 1996. Formal Ethics. London: Routledge. Gensler, New York: Oxford University ed. 1987. Moral Dilemmas. Press. Christopher, In Studies in the Way of Words. Cambridge, Grice, Paul. 1989. "Logic and Conversation." Harvard University Press. Press. Hare, R. M. 1963. Freedom and Reason. New York: Oxford University and Belief. Ithaca, N.Y.: Cornell University Press. Jaakko. 1962. Knowledge Hintikka, Gowans, Horty, vol. Mass.: John. 1994. "Moral Dilemmas 23, pp. 35-65. and Non-monotonic Logic." Journal of Philosophical Logic, Kant, Kekes, vol. Immanuel. John. 34, pp. 1984. 1797. The Metaphysical of Virtue. Excerpts reprinted in Gowans. Principles " Can' and Two Kinds of Morality." Philosophical Quarterly, 'Ought Implies Philosophical Review, vol. 71, pp. 139-58. Reprinted 459-67. Lemmon, in Gowans. E. J. 1962. "Moral Dilemmas." and In Semantics of Natural Language, ed. Davidson 1972. "General Semantics." Lewis, David. Harman. Dordrecht: Reidel. vol. 77, and Consistency." Journal of Philosophy, Marcus, Ruth Barcan. 1980. "Moral Dilemmas in Gowans. 121-36. Reprinted pp. in Linguistic in a Grammar." In Universals James. 1968. "The Role of Semantics McCawley, Theory, ed. Bach and Harms. New York: Holt, Rinehart, and Winston. 1968. "On Second Order Norms." Ratio, vol. 10, pp. 101-15. Moritz, Manfred. 1990. "Ought Implies Can: Erasmus Luther and R. M. Hare." Charles. Pigden, pp. 2-30. Sophia, vol. 29, Plantinga, Alvin. 1974. God, Freedom, and Evil. New York: Harper and Row. OUGHTDOES NOT IMPLYCAN I 105 1987. In Contradiction. Dordrecht: Martinus Nijhoff. Priest, Graham. 1987. Ethical Idealism. Berkeley: University of California Rescher, Nicholas. Sapontzis, 383-93. view, vol. _. _. Steve. 1991. "Ought Does 1984. to Have' Dilemmas. Imply Can.99 Southern 'Ought' Conversationally Journal Implies vol. of Philosophy, Press. Chapter 2. vol. 29, pp. Re? Sinnott-Armstrong, 93, 1985. 1988a. pp. " Walter. 249-62. 'Ought Moral " 'Can'." Philosophical 45, pp. 44-48. and 'Could Have'." Blackwell. Analysis, London: _. Stockwell, of English. van Fraassen, Bas. 1973. "Values and the Heart's Command." Journal of Philosophy, vol. 70, pp. 5-19. Reprinted in Gowans. von Wright, Georg Henrik. 1963. Norm and Action. London: Routledge. Morton. 1981. What Is and What Ought to be Done. New York: Oxford. White, Michael. 1996. The Concept of Moral Obligation. Zimmerman, Cambridge: Cambridge Uni? versity Press. 1988b. "Promises Which Cannot Be Kept." Philosophia, vol. Robert; Paul Schachter; and Barbara Partee. 1973. The Major New York: Holt, Rinehart and Winston. 18, pp. 399-407. Syntactic Structures