I OUGHT THEREFORE I CAN OBEY Peter B

I OUGHT, THEREFORE I CAN OBEY Peter B. M. Vranas vranas@wisc. edu University of Wisconsin-Madison Facts & Norms III Workshop, 24 August 2017

OUGHT IMPLIES CAN (OIC) By virtue of conceptual necessity, if an agent at a given time ought (i. e. , has an objective, pro tanto obligation) to do something, then the agent at that time can (i. e. , has both the ability and the opportunity to) do the thing. l Obligations, abilities, and opportunities are indexed to times. OIC relates obligations at a given time with abilities and opportunities at the same time. l

TWO LIMITATIONS OF OIC explains why we don’t have certain infeasible obligations (i. e. , obligations we can’t obey). l But OIC doesn’t explain why we don’t have infeasible obligations to feel or infeasible conditional obligations. l Primary project: generalize OIC to reach a principle informative about all kinds of obligations. l Secondary project: defend an ought-implies-canavoid principle. l

OVERVIEW Part 1 UNCONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN SATISFY (OICS) Part 2 CONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN OBEY (OICO) Part 3 OUGHT IMPLIES LIVE OPTION (OILO)

OIC REFORMULATED Satisfaction proposition of your obligation to help me: the proposition that you help me. l S at t can satisfy an obligation iff S at t can cause the satisfaction proposition of the obligation to be true (i. e. , can actualize that proposition). l (OIC) If S at t has an obligation to do something, then S at t can do the thing. Equivalently: l (OIC) If S at t has an obligation to do something, then S at t can satisfy the obligation (i. e. , can actualize its satisfaction proposition). l

NON-AGENTIAL OBLIGATIONS Agential obligation (i. e. , obligation to do): obligation to make something the case. Are there non-agential obligations? l Obligation to attend my wedding. If you are brought to my wedding by force, you attend but you don’t make it the case that you attend. l Obligation not to kill me. If I kill you, you don’t kill me but you don’t make it the case that you don’t kill me. l

OUGHT IMPLIES CAN SATISFY If you cannot attend my wedding, then you lack both (1) the agential obligation to make it the case that you attend and (2) the non-agential obligation to attend. But OIC doesn’t explain why you lack (2). l (OIC) If S at t has an obligation to do something, then S at t can satisfy the obligation. l (OICS) If S at t has an unconditional obligation, then S at t can satisfy the obligation. l OICS is restricted to unconditional obligations. l

LOSING OBLIGATIONS OICS explains why, if you have an unconditional obligation and you become unable to satisfy it, then you lose it. But in some cases you lose an obligation without becoming unable to satisfy it. l You promise that my dog will be dead by midnight. When you poison the dog, you are done with your obligation, so you lose it. But you can still satisfy it: you can kill the dog by shooting it. l The point is not that OICS is false, but that it doesn’t explain why you lose the obligation. l

OUGHT IMPLIES CAN VIOLATE? One might propose that you lose the obligation because you become unable to violate it. l (OICV) If S at t has an unconditional obligation, then S at t can violate the obligation. l Counterexample to OICV: You promise that your daughter will not participate in a study. You can satisfy your obligation (by withholding your consent) but you cannot violate it (you cannot make it the case that she consents). l

ABILITY-NECESSITY A proposition is inevitable (i. e. , ability-necessary) for S at t iff it would be true no matter what S were to do, among the things that at t S can do. l “The sun will rise tomorrow” is inevitable for you now. “The sun will rise tomorrow and you will not scratch your nose in the next minute” is not inevitable for you now but will be inevitable for you in a minute if you don’t scratch your nose. l P is ability-possible for S at t iff it might be true if S were to do something that at t S can do. l

OUGHT IMPLIES POSSIBLE VIOLATION (OIPV) If S at t has an obligation, then its violation proposition is ability-possible for S at t. l OIPV escapes the counterexample to OICV: you cannot make it the case that your daughter participates in the study, but it is ability-possible for you that she participates. l OIPV (but not OICS) explains why you lose the obligation in the dog example. OICS (but not OIPV) explains why you have no obligation to win the lottery. l

PART 2 Part 1 UNCONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN SATISFY (OICS) Part 2 CONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN OBEY (OICO) Part 3 OUGHT IMPLIES LIVE OPTION (OILO)

CONDITIONAL OBLIGATIONS You promise me that, if (F) your flight is delayed, (C) you will call me and tell me so. As a result, you acquire a conditional obligation O. l O is violated iff F & ~C is true. Is O satisfied iff it is not violated (i. e. , if ~F C is true)? No: if ~F & C (and thus ~F C) is true, O is not satisfied. O is satisfied instead iff F & C is true. l O is neither satisfied nor violated (i. e. , is avoided) iff F is false. An obligation is unconditional iff its satisfaction proposition is the negation of its violation proposition. l

THE FAILURE OF OICS* (OICS*) If S at t has a conditional obligation, then S at t can satisfy the obligation. l Counterexample to OICS*: You have a conditional obligation to hide in the basement if there is a tornado, but you cannot make it the case that there is a tornado and you hide. l No counterexample to OIPV: It is ability-possible for you that there is a tornado and you don’t hide. l But OIPV doesn’t explain why you have no conditional obligation to win the lottery if I don’t: it is ability-possible for you that neither of us wins. l

OUGHT IMPLIES CAN OBEY An obligation is obeyed iff it is not violated (i. e. , iff it is either satisfied or avoided). Obedience proposition: negation of violation proposition. l Example: consider your obligation to (C) confess if (G) you are guilty. Its obedience proposition is ~(G & ~C); i. e. , ~G C; i. e. , G C. l For unconditional obligations, avoidance is impossible, so obedience amounts to satisfaction. l (OICO) If S at t has an obligation, then S at t can obey the obligation (i. e. , can actualize its obedience proposition). [OICO entails OICS. ] l

IN SUPPORT OF OICO escapes the counterexample to OICS*: you can make it the case that there is no tornado or you hide. OICO explains what OIPV doesn’t: you can’t make it the case that I win the lottery or you win. l Argument for OICO: (1) If S at t has an obligation, its violation proposition is impermissible for S at t. (2) If the violation proposition V of an obligation is impermissible for S at t, then S at t can prevent V (i. e. , can actualize ~V). So: If S at t has an obligation, S at t can prevent its violation proposition. l

WHY OIPV IS STILL NEEDED OICO doesn’t always explain why agents lose conditional obligations. l You promise that my dog will be dead by midnight if it is alive at 1 pm. When you poison the dog, you lose your conditional obligation. But you can still actualize its obedience proposition (i. e. , that the dog is—not alive at 1 pm or—dead by midnight): you can kill the dog by shooting it. l OIPV explains why you lose the obligation: its obedience proposition becomes inevitable for you. l

PART 3 Part 1 UNCONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN SATISFY (OICS) Part 2 CONDITIONAL OBLIGATIONS: OUGHT IMPLIES CAN OBEY (OICO) Part 3 OUGHT IMPLIES LIVE OPTION (OILO)

COMBINING OICO WITH OIPV (OICO) If S at t has an obligation, then S at t can actualize its obedience proposition. l (OICO) If S at t has an obligation, then its obedience proposition is feasible for S at t. l (OIPV) If S at t has an obligation, then its violation proposition is ability-possible for S at t. l (OIPV) If S at t has an obligation, then its obedience proposition is not inevitable for S at t. l (OICO & OIPC) If S at t has an obligation, then its obedience proposition is feasible but not inevitable for S at t. l

OUGHT IMPLIES LIVE OPTION Inevitable Infeasible Non-options The sun rises tomorrow Feasible Dead options You run today or the sun rises tomorrow Not inevitable Non-options You win the lottery Live options You run today (OILO) If S at t has an obligation, then its obedience proposition is a live option (i. e. , is feasible but not inevitable) for S at t.

CONTROLLABILITY P is controllable iff both P and ~P are feasible. l Alternative proposal: If S at t has an obligation, its obedience proposition is controllable by S at t. l Inevitable Not inevitable Infeasible Uncontrollable Feasible Uncontrollable live options Your daughter does not participate in the study Controllable live options You run today

CONCLUSION Left for the discussion period - Is OILO the whole story? - OILO versus obligations partly about the past. - OILO versus epistemic obligations. l Methodology formulating and assessing ought -implies-can principles - Formulate them in terms of obligations, not in terms of what an agent is obligated to do. - Assess them in terms of explanatory power (to explain why agents lack or lose obligations). l

OBLIGATIONS ABOUT THE PAST You acquire an obligation to pay all installments of a loan on time (on 1/5, 2/5, …). You pay on 1/5 but not by 2/5. OIC explains why you lose the obligation, but only if you lose it on 2/5. Then on 2/4 you have an obligation partly about the past. l This is possible: if at 8 am you promise to never smoke again, at any time shortly after 8 am you have an obligation partly about the past. l By OIC, (1) on 2/4 you can pay all installments on time, so (2) on 2/4 you can pay the January installment on time. Reply: (at t you can actualize P and P entails P ) ⇏ at t you can actualize P. l

EPISTEMIC OBLIGATIONS Objection to OICS: (1) You have an epistemic obligation to start believing P, but (2) you can’t start believing P (you can’t believe P at will). l Reply: (1) ⇒You have evidence for P. l Response: Even if you have evidence for P, you can’t make it the case that you start believing P. l Reply: You add numbers and find 987, but suspend belief until you check. You find again 987, and decide to stop checking and start believing that the sum is 987. You made it the case that you started believing. l

IS OILO THE WHOLE STORY? Apparently, OILO doesn’t explain why you have no obligation to (R) run tomorrow if (P) you become paralyzed tonight: ~P R is feasible (since you can actualize R) but not inevitable (it might be false if P were true) for you. l Reply: Inevitably, you won’t—and thus you can’t —actualize ~P R: (1) if P were false, ~P R would become settled tonight, but (2) if P were true, ~P R would be false (so in either case you wouldn’t actualize it). l

GRAHAM’S OBJECTION TO OIC Unless a janitor J kills her, a surgeon S will kill an innocent man M to save ten people. l Graham claims: The best explanation of (1) is (2) even if S cannot refrain from killing M: (1) It is morally permissible for J to kill S. (2) If J had not killed S, S would have morally impermissibly killed M. [So: (3) S has a moral obligation not to kill M—contrary to OIC. ] l My reply: Replace S with a robot R. Then: (1 ) It is permissible for J to destroy R. But (2 ) cannot explain this, since R has no obligations. I propose instead: (2*) It is impermissible that S kills M. l