H. P. GRICE E J. L. SPERANZA: LA CONVERSAZIONE -- I VERBALI: PADOVANI

 G.: Let us begin with the slogan in its vulgar splendour: il fine giustifica i mezzi. S.: A sentence that has done more work after dinner than before reflection. G.: Precisely. And Padovani, being a neo-scholastic in wartime print, wishes to know whether it is true, false, or merely badly bred. S.: Probably all three, in alternating moods. G.: Let us try to improve its manners by formalisation. Let f stand for fine, the end, and m for mezzo, the means. S.: And V for volere. G.: Yes. Then the first temptation is Aristotelian and schoolboyish: if the agent wills f, and m is necessary to f, then he wills m. S.: So one writes something like V(f), and if m is a necessary means to f, then V(m). G.: Exactly. The old maxim: he who wills the end wills the means. S.: Which is often true, except when it is merely hopeful. G.: Quite. Means-end rationality is not a miracle-worker. It tells us something about consistency within willing, not yet anything about justification. S.: So the formula gives us transmission of volition, not moral vindication. G.: Precisely. The vulgar slogan, however, uses giustifica, and that is the troublesome word. S.: Because giustifica is not the same as “entails a further willing.” G.: Exactly. If I will f and therefore will m, it hardly follows that m is justified merely because it lies on the route. S.: Otherwise every scoundrel with a timetable would count as a moral theorist. G.: A fair summary of several traditions. So let us mark the first-order case. V(f) and N(m,f), where N expresses that m is a necessary means to f. S.: Then by practical rationality one may derive V(m). G.: Yes. But nothing yet deserving J. S.: Then J, giustifica, cannot be reduced to V at the first order. G.: That is my proposal. J must be treated as a higher-order buletic operator. S.: Meaning that it ranges over volitions rather than over bare states of affairs. G.: Exactly. It does not simply attach to m or f as objects. It attaches to willings qua willings. S.: So not J(m,f), but something like J(V(m),V(f)). G.: Better still, J may itself be definable in terms of a second-order willing. S.: A willing of a willing. G.: Precisely. Something Prichardian in its awkwardness, and Kantian in its ambition. The agent not only wills f and thereby wills m; he wills that this willing be the sort of willing he can own. S.: Which already sounds like trouble for Machiavelli. G.: Trouble is the beginning of philosophy. S.: Then state the proposal cleanly. G.: Very well. First-order means-end rationality gives: If V(f) and N(m,f), then rational pressure towards V(m). S.: Pressure, not yet legitimacy. G.: Exactly. Now let J(V(m)) mean: the agent wills his willing of m under a higher-order endorsement. S.: So he does not merely will m, but wills that he will m. G.: Yes. Or, if one prefers, he reflectively ratifies the willing of m. S.: And similarly perhaps for f. G.: Necessarily. For if the end itself is not reflectively ratified, then the chain is rotten from the top. S.: So one needs J(V(f)) as well. G.: Indeed. The vulgar slogan starts from the end as if the end arrived with a halo attached. Padovani, being scholastic enough to distrust halos, wants to ask what sort of end could justify anything. S.: Then means-end rationality is subordinate to end-criticism. G.: Precisely. And end-criticism, in our formalism, becomes criticism of the willing of the end. S.: So the agent is free not merely in willing, but in taking a stand on what he wills to will. G.: Very good. That is the crucial turn. Freedom enters not as random spontaneity but as higher-order buletic governance. S.: A man may will revenge. That is first-order enough. The question is whether he can will that he will revenge. G.: Exactly. And if he cannot stably or lucidly endorse that willing, the mere fact that revenge has convenient means does not save it. S.: Then J is not doxastic. G.: Certainly not. It is not “I believe this willing to be justified.” Belief alone is too cheap. J belongs to the order of volitional self-appropriation. S.: So one might define: J(V(x)) iff V(V(x)) under conditions of reflective freedom. G.: Yes, with the rider that the second-order willing is not a mere repetition but an endorsement. S.: Otherwise obsession would count as morality. G.: And many deans would become saints. S.: A painful possibility. G.: Let us avoid it. So perhaps: J(V(x)) =df the agent freely wills that he will x, and can sustain that willing under universal practical scrutiny. S.: You have smuggled Kant in through “universal.” G.: Deliberately. Padovani wants Machiavelli disciplined by a classical and Catholic moral framework, but we may let Kant assist with the policing. S.: Then the practical syllogism is not enough. G.: Never was. Aristotle gives one a route from appetite or wish to action under some conception of the good. He does not by that alone answer whether the conception itself is fit to legislate for a free rational agent. S.: So our hierarchy is this. First-order: V(f). N(m,f). Therefore V(m). G.: Yes. Second-order: V(V(f)). V(V(m)). Or more carefully, the agent endorses willing f and willing m. S.: And J is the name for that endorsed willing. G.: Exactly. J(V(m)) holds only if V(m) is itself willed as a willing under a higher-order act. S.: Then the slogan “the end justifies the means” becomes something like: J(V(f)) and N(m,f) may yield J(V(m)). G.: Better. But only “may.” One must not let necessity of means smuggle in automatic justification. S.: Because the means may introduce a fresh moral defect. G.: Precisely. Suppose f is some allowable end, but m involves lying, cruelty, or murder. The higher-order endorsement of V(f) does not simply trickle down like holy oil. S.: Then one needs a further condition that V(m) itself be endorsable. G.: Exactly. So: J(V(m)) iff N(m,f) and J(V(f)) and E(V(m)), where E marks higher-order endorsability of the means-willing itself. S.: Which is very nearly to deny the slogan. G.: Or to civilise it into near-unrecognisability, which is often the charitable way to deny a slogan. S.: Padovani in 1917 would have approved the charity and the denial. G.: With some reservations from the editor, no doubt. S.: As indeed the journal suggests. G.: Quite. Now let us test the machinery on a simple case: veracity. S.: A Kantian delight. G.: Also a scholastic headache. Take the commandment against false witness, or more broadly the duty of truthfulness. Suppose f is the end of preserving a friend from danger. S.: And m is lying to a murderer at the door. G.: The common undergraduate begins at once to feel important. S.: As undergraduates do when murderers are introduced. G.: Quite. So first-order practical rationality says: V(f), preserve the friend. N(m,f), the lie is the necessary means. Therefore V(m), lie. S.: Means-end rationality delivers the lie without blushing. G.: Exactly. But Padovani wants to ask whether J(V(m)) follows. S.: Kant says no, or nearly no, because the maxim of lying cannot be universally legislated. G.: Yes. The will that wills itself rationally cannot endorse the willing-to-lie as such without damaging the kingdom of ends. S.: Because others are then treated as instruments of one’s management of appearances. G.: Precisely. The categorical imperative enters as the condition under which higher-order willing counts as justified rather than merely reflective. S.: So a maxim of prudence is not yet a categorical principle. G.: That is the point. Counsels of prudence tell one how to get what one happens to want. The categorical imperative tells one what sort of willing can be owned by a free rational being among other such beings. S.: Then the means-end chain lives entirely below the level of final justification. G.: Exactly. It is necessary for rational agency, but insufficient for moral agency. S.: So Machiavelli thrives in the lower level. G.: A neat way of putting it. Machiavelli is often strongest where one is discussing efficacy under given political ends. Padovani wants to ask whether efficacy can ever by itself become justification. S.: And your answer is: only if one mistakes first-order coherence for second-order endorsement. G.: Precisely. The prince may will stability, and will cruelty as a means, and do so with magnificent consistency. That gives him practical unity, not moral legitimacy. S.: Then J must range over maxims, not just isolated acts. G.: Better still. For what the agent wills to will is often not a token action but a kind of action under a description. S.: So J(V(m)) is really shorthand for endorsement of a maxim containing m. G.: Yes. For example: I will that, when political stability requires deception, I deceive. S.: Which the kingdom of ends may find indecorous. G.: It tends to. Then our formula must notice descriptions. The same bodily motion can be described as preserving order, or as murdering rivals. S.: Which means that higher-order willing is sensitive to the specification of the object willed. G.: Very much so. One does not justify a bare event, but a willing under a rational description. S.: Then the slogan should be rewritten: A reflectively justified willing of an end does not by itself justify every willing of every necessary means; only those means-willings that can themselves be reflectively and universally endorsed are justified. G.: Excellent. Not fit for a banner, but fit for philosophy. S.: So Padovani’s question from 1917 receives a dry answer: no, unless by “giustifica” one mean something stronger than means-end rationality, in which case usually still no. G.: Splendid. Keep the “usually.” S.: Because one should not become melodramatic. G.: Or journalistic. Now let us consider whether second-order willing is enough. You may object that a fanatic can will his own willing all the way up. S.: Easily. He may V(f), V(V(f)), V(V(V(f)))), and so on, until the notation itself loses faith. G.: Exactly. Infinite access to one’s own willing does not guarantee moral success. S.: So the higher-order structure is necessary but not sufficient. G.: Correct. One also needs a test of the content of the maxim. S.: Hence Kant. G.: Hence Kant, and perhaps Prichard’s reminder that duty is not reducible to what one happens strongly or reflectively to want. S.: Then J must include not only higher-order willing but a norm on higher-order willing. G.: Yes. Call it U, for universalizability, if you like. S.: Then: J(V(x)) iff V(V(x)) and U(x). G.: Better: J(V(x)) iff the agent freely endorses V(x) under a maxim fit for universal legislation among ends in a kingdom of ends. S.: Which sounds better in German than in English. G.: Most police do. S.: Then Machiavelli is refuted, or not, depending on whether his maxims survive U. G.: Quite. Some prudential maxims may survive in a restricted political form. Others collapse at once because they require asymmetry: I may deceive, others may not deceive me. S.: The old pleasure of universalisation. G.: Indeed. It ruins many careers. S.: Then Padovani, as neo-scholastic, might not phrase it in Kantian terms, but he would agree that an evil end cannot sanctify an evil means, and a merely useful end cannot baptise moral defect. G.: Precisely. We are letting Kant and Prichard lend him some English and German machinery. S.: And Grice some symbolic tidiness. G.: Such as it is. Now, one further distinction. There is “willing m because m is necessary to f,” and there is “willing to be the sort of person who wills m under that description.” S.: The latter is the real higher-order burden. G.: Exactly. It is one thing to will a lie in panic. Another to will oneself as a liar under a principle. S.: That does sharpen the conscience. G.: Philosophy occasionally has that use. S.: Then the commandment case becomes especially instructive. If truthfulness belongs to the conditions of mutual respect in the kingdom of ends, then a lying means threatens the very order within which justification is sought. G.: Precisely. The means may damage the medium of justification itself. S.: Which is rather elegant. G.: Dryly so. If a regime of willing depends upon mutual recognisability of rational beings, then means that systematically exploit or degrade that recognisability attack the conditions of J. S.: So some means are self-undermining relative to higher-order justification. G.: Exactly. Torture is a good grim example. One may will political security, and will torture as a means, but the higher-order endorsement required for J is corrupted because the means destroys the standing of persons as ends. S.: Then the slogan fails not only morally but architecturally. G.: Very good. It confuses the lower architecture of efficacy with the higher architecture of justifiable willing. S.: So “the end justifies the means” is, in your formal reconstruction, an equivocation on levels. G.: Exactly. It slides from: The end explains why the means is chosen, to: The end morally licenses the means. S.: And the slide is illicit. G.: Entirely. Explanation is not justification. S.: Nor is coherence endorsement. G.: Nor is endorsement universal lawfulness. S.: We are climbing nicely. G.: Philosophy is mostly stairs badly lit. S.: Then let us descend to Padovani again. In 1915 he writes on Spinoza; in 1917 he asks this wartime question. One can see why the period would sharpen the distinction between ends and retrospective excuse. G.: Yes. War is where slogans about ends and means become indecently practical. S.: So Grice’s reminiscence of his father and the Great War gives the whole thing more than seminar charm. G.: Quite. The question is not merely whether a prince may deceive, but whether collective suffering is ever “justified” by ends proclaimed after the dead have done the hard part. S.: Which suggests that “justification” is often retrospective rhetoric. G.: Often. One says afterwards that the end justified the means because one dislikes admitting that the means have occurred without moral redemption. S.: Burial as argument. G.: Very good. Keep that too. S.: You are becoming positively distributive. G.: Do not abuse the occasion. Now, would you say J is iterated V all the way up? S.: Not quite. I would say J is definable by a hierarchy of V, but constrained by a non-buletic norm, call it U, or kingdom-of-ends fitness. G.: Yes. Otherwise the fanatic again reappears, infinitely reflective and infinitely appalling. S.: So our final schema might be: If V(f) and N(m,f), then rational pressure towards V(m). But J(V(m)) only if J(V(f)) and the maxim containing m is itself fit for higher-order endorsement under universal practical law. G.: Excellent. Which means the end does not justify the means merely by being the end. S.: It may at most contribute to the intelligibility of the means. G.: Precisely. The means becomes explicable, perhaps prudent, perhaps even unavoidable. But justification requires another tribunal. S.: And that tribunal is the free agent’s higher-order willing under universalisable maxims. G.: Very good. Padovani would perhaps prefer a more scholastic tribunal, but he would recognise the need for one above prudence. S.: Then the slogan is really two questions disguised as one: Does the end require the means? And: May the agent endorse himself in willing that means for that end? G.: Exactly. The first is practical reasoning. The second is moral philosophy. S.: And Machiavelli is strongest in the first, weakest in the second. G.: That is a justly English verdict. S.: Then perhaps one may answer Padovani in one sentence: The end may determine the means as matter of prudence, but only a higher-order willing, answerable to universal practical reason, can justify the willing of the means. G.: Admirable. Too long for a slogan, which is why it is safer. S.: Dry enough? G.: Sufficiently neo-scholastic, with a little Königsberg hidden in the sacristy.