How to Motivate the Maxim that “Ought” Implies “Can” to Defend the Principle of Alternate Possibilities

By Sean Armil

Winner of the Gerritt and Edith Schipper Undergraduate Award for Outstanding Undergraduate Paper at the 54th Annual Meeting of the Florida Philosophical Association

Sean Armil, University of Florida

In this paper, I will examine John Martin Fischer’s rejection of the Maxim that “ought” implies “can.” He is motivated to reject the Maxim because of an argument, presented by David Copp, which uses the Maxim to derive the Principle of Alternate Possibilities (PAP) with respect to blameworthiness. PAP with respect to blameworthiness states that people are not morally blameworthy for what they have done unless they could have acted otherwise. Fischer is a semi- compatibilist; he believes that determinism and the ability to do otherwise than one actually does are incompatible, but that determinism and moral responsibility are compatible. Thus, Fischer believes that PAP is false and counters Copp’s derivation of PAP by arguing against the two motivations Copp provides for the Maxim. Copp’s first motivation is an argument based on agent requirements and fairness. Fischer counters this argument using a Frankfurt-style example. Copp’s second motivation is an argument which appeals to the purpose of moral requirements as being action guiding. Fischer believes this argument is flawed because it ignores a distinction between metaphysical and epistemic options.

I will argue that the Frankfurt-style example Fischer provides is not enough to counter Copp’s first motivation. I will also argue that the key to rescuing the second motivation from its equivocation between metaphysical and epistemic options lies in Copp’s compatibilist analysis of “can.” I think that Copp’s compatibilist position faces a serious objection, though, so I will suggest a way to revise his compatibilism to meet this objection in a way that is sympathetic to Fischer’s semi- compatibilist position.

Determinism is the view that the laws of nature, combined with a complete description of the facts of the universe at time T, entail every fact about the universe after time T. Determinism seems to threaten our control over our actions and our conception of free will that involves alternate possibilities. If the laws of nature and facts of the universe before we were born determine our actions, then it seems that we cannot do anything other than what the laws of nature and facts of the universe determine us to do. This belief that determinism entails a lack of alternate possibilities motivates a common sense argument which purports to show that determinism is incompatible with moral responsibility. That argument can be stated like this:

  1. People are not morallyr esponsible for what they have done unless they could have acted otherwise.
  2. In a deterministic universe, people could not have acted otherwise than they actually did.
  3. Therefore, no one is morally responsible in a deterministic universe.

The first premise of this argument is the intuitive but controversial Principle of Alternate Possibilities. PAP seems to play an important role in our moral deliberations. For example, we do not hold a trolley driver morally responsible for running over someone tied to the track if the trolley brake is broken, because the driver could not have done otherwise than run over the person. However, if the brake is functioning fine and the trolley runs over a person because the driver was not paying attention, then we would hold him morally responsible. He could have paid attention, stopped his trolley, and not run the person over. It seems that we are assigning blame and responsibility in these cases based on the trolley driver’s ability to act otherwise than he did.

Harry Frankfurt famously argued that PAP is false and that the ability to do otherwise is not a necessary condition for holding a person morally responsible. He developed a series of examples which elicit intuitions inclining us to hold a person morally responsible for an action, even though that person could not have done otherwise. The examples work something like this: A neurosurgeon named Black vows to intervene to ensure a person does X only if the person is not going to do X. Black will intervene through a device implanted in the person’s head that detects when she is about to avoid doing X, and then zaps her neurons somehow so that she has to do X. However, if the person is going to do X on her own, the device will not go off. Even though the person cannot do otherwise than X (because if she tries not to do X, Black’s device will intervene and force her to do X), we would still want to hold her morally responsible for dong X if she does. If she does X, she does what she wanted to do without anyone forcing her to do it. It seems that, despite her lack of alternate possibilities, we would want to hold her responsible when Black does not intervene. Frankfurt argues that these examples show PAP is false. Frankfurt’s position is controversial, and many have argued against it.

David Copp has argued that rejecting PAP comes at a high price; if we reject PAP, we will have to reject the Maxim that “ought implies can.” Copp believes that, since PAP and the Maxim are inextricably linked and the Maxim has more intuitive support than do Frankfurt examples, we should opt to keep PAP. The Maxim states that “(a) A person is all-in morally required to do A in a situation only if she can do A in that situation and (b) a person is all-in morally required not to do A in a situation only if she can do something other than A in that situation.”1 Copp argues that PAP can be derived from the Maxim in the following way:

  1. S is morally blameworthy for doing A only if S was all-in morallyr equired not to A.
  2. S was all-in morally required not to do A only if S could have done something other than A.2
  3. Therefore, S is morally blameworthy for doing A only if S could have done something other than A (PAP).3

I should note that Frankfurt’s version of PAP involves moral responsibility and that the version Copp just derived only involves blameworthiness. Frankfurt thought that his version of PAP implied Copp’s version, though, and since Copp’s argument derives PAP with respect to blameworthiness, I will discuss PAP with respect to blameworthiness for the rest of the paper.

Since the Maxim and PAP are linked in such a close way, motivations for the Maxim can also be counted as support for PAP. Copp provides two motivations for accepting the Maxim: the argument from fairness and the argument from the point of moral requirements. The argument from fairness is based on our conception of agent-requirements. We intuitively think it would be unfair for a teacher to require her students to read and comprehend an entire book in one night if the students were not capable of such a thing. To spell out this intuition more precisely, we think:

It is unfair for one person to require that another person do some X, if the latter person cannot do X.

This is analogous to the following, which supports the Maxim:
It is unfair for morality to require that a person do some X if the person cannot do X.

The second motivation for the Maxim is an argument based on the action-guiding nature of moral requirements. After considering several different ways to approach this point, Copp puts his argument as follows:

  1. If an agent is morally required to do A in a particular situation, then all other options she faces are morally ruled out.
  2. If the agent cannot do A, then doing A is not among her options.
  3. Hence, if an agent is morally required to do A but cannot do A, then all of her options are morally ruled out.
  4. Information that an agent is morally required to do something provides her with guidance among her options by distinguishing between options that are morally ruled out and options that are not morally ruled out.
  5. But, if all the agent’s options are morally ruled out, the moral requirement has not served its purpose by providing the agent with guidance.4

In order for moral requirements to be action-guiding, it must be the case that the agent can do what the moral requirement requires.

John Martin Fischer argues that Copp’s motivations for the Maxim are not strong enough. He thinks we should reject the Maxim along with PAP. He first counters the arguments from fairness by questioning the intuitions supporting premise 1. Fischer argues that we could intuitively blame someone for failing to do X even when the person could not do X. Fischer agrees with Copp that it would be unfair to blame an agent for some omission, such as failing to read and comprehend an entire book in one night, that the agent could not help but commit. But he thinks Copp’s examples only cover a subset of omissions called “complex omissions.” He thinks that a critique of the argument from fairness can be drawn from another subset of omissions called “simple omissions.” He argues that an agent can be blamed for simple omissions, like failing to move his body, even when the agent could not have done otherwise. To motivate this claim, he uses a Frankfurt example in which we should intuitively judge that it would be fair to blame a person for not moving his body, even though he could not have done otherwise.

In this example, a man named Stanley does not move his body at all. Some counterfactual intervener has a vested interest in keeping Stanley from moving, and will strike him with paralysis if he shows any inclination to move. If Stanley does not move at all, the intervener will not intervene. It seems that we could intuitively blame Stanley for not moving, even though he lacks the ability to move. Fischer thinks this example is one of many in which an agent could be blamed for failing to do X, even though he could not do X. I do not think this is a successful counterexample to the argument from fairness, because I disagree that we would hold Stanley responsible for failing to move. I plan to criticize the Stanley example later with a strategy similar to the one Fischer employs against Copp’s argument from the point of moral requirements. So now I will spell out Fischer’s strategy to counter that argument.

Fischer contends that Copp’s argument based on the point of moral requirements “infelicitously elides the distinction between metaphysical and epistemic options.”5 Fischer thinks that Copp’s argument is invalid because it equivocates between two different types of options: metaphysical and epistemic. Metaphysical options are courses genuinely available to the agent, while epistemic options are courses that are available to the agent for all the agent knows. Fischer thinks that, in order to avoid equivocation and properly reflect the metaphysical/epistemic distinction, the first two premises of the argument should be stated like this:

  1. If an agent is morally required to do A in a particular situation, then all other epistemic options she faces are morally ruled out.
  2. If an agent cannot do A, then doing A is not among her metaphysical options.6

Fischer views this reformulated argument as obviously flawed. If an agent lacks a metaphysical option, it does not follow that she lacks the corresponding epistemic option. If an agent cannot A, it is not necessarily true that the agent knows this. So now let me return to my response to Fischer’s critique of the argument from fairness.

I agree that an agent can lack a metaphysical option but still have the corresponding epistemic option. I think that Fischer infelicitously elided this distinction in his criticism of the argument from fairness. In the Frankfurt example, Stanley had the epistemic option of standing up (since for all he knows, he can), but lacked the metaphysical option. Trying to act on this epistemic option is one of Stanley’s alternate possibilities. I think that Stanley should be held responsible specifically because of his failure to consider this epistemic option, and not because of his failure to move.

Fischer could respond to this by saying that the intervener’s actions are such that Stanley actually lacks the epistemic option. Stanley’s intervener could have put a neural implant in Stanley like Black did in the first Frankfurt example. Stanley’s implant will intervene and zap Stanley if his neurons show any inclination that he was considering moving. Stanley’s intervener could be an advanced neurosurgeon who is able to tell which neural event is a sure sign that Stanley will consider moving as an epistemic option. Once this sign is observed, the implant could zap Stanley so that he is paralyzed and cannot move. If there is some neural event which is a sure sign of Stanley’s consideration, this means that Stanley’s mental process is deterministic. Many libertarians, such as David Widerker, see this assumption of determinism as problematic, since they reject that we could be responsible based on such deterministic processes.7 However, I see the assumption of determinism as problematic for a different reason.

If the sign in Stanley’s brain is the beginning of a deterministic process that leads to the consideration of his epistemic option to move, then I think the sign is part of the consideration. The sign, whatever type of neural event it is, would mark the beginning of Stanley’s consideration of his epistemic option to move his body. The intervener’s implant could stop Stanley before the process of consideration completes, but cannot stop the process from beginning. Even though Stanley ends up paralyzed and never gets to try to act on this epistemic option, this route still represents an alternate possibility. I think it is intuitive to blame Stanley for his failure to explore this possibility and try to get up. I will spell out this intuition that Stanley is responsible for not trying to stand to show that Fischer has not provided an example in which it is fair to blame a person for not doing X when the person could have done X, and thus that his example does not affect the argument from fairness.

Suppose there are two different Stanleys, and each sees a child in front of him drowning. Neither of them has a genuine metaphysical option of rescuing the child because of the intervener and the neural implant. But neither knows the intervener will paralyze them, so each Stanley has available to him an epistemic option of getting up to save the child. Both should at least consider whether they should try to save the child. If one of them does not begin this consideration, we would hold him responsible for not doing so. Suppose Stanley-1 does not begin considering whether to save the child, and just remains seated while the child drowns. I think we should view him as morally blameworthy for failing to try to save the child because he did not begin this epistemic consideration. He has some sort of all-in moral requirement to do what he believes he can to attempt to save the child.

His counterpart, Stanley-2, became inclined to get up and go save the child, but was then paralyzed by the intervener’s implant. We do not have to blame Stanley-1 or Stanley-2 for failing to save the child (because neither could) but it still seems like we would want to make some sort of moral distinction between them. We intuitively find Stanley-1 blameworthy, and I think we should blame him for callous indifference to the child’s fate. We know Stanley-2 does not share this indifference because he at least considered saving the child and fulfilled his duty to do what he believes he can. It is fair to require Stanley-1 to begin considering to save the child, because this is something Stanley-1 could have done. Thus, I think Fischer has failed to provide an example of a situation where we should blame someone for failing to do X, if he could not do X. We can still blame Stanley for failing to do something he could have done. Thus, Copp’s argument from fairness still counts as a successful motivation for the Maxim.

My plan now is to examine Fischer’s criticism of the argument from the point of moral requirements. Remember, Fischer thought the argument equivocated between metaphysical and epistemic options, and should be properly stated as follows:

  1. If an agent is morally required to do A in a particular situation, then all other epistemic options she faces are morally ruled out.
  2. If an agent cannot do A, then doing A is not among her metaphysical options.

It seems like the argument, to avoid equivocation, has to deal consistently with either metaphysical or epistemic options. Fischer thinks that “moral requirements insert themselves into the space of epistemic possibilities, not directly into the space of metaphysical possibilities.”8 So, if the argument from the point of moral requirements is stated in terms of epistemic options, it goes as follows:

  1. If an agent is morally required to do A in a particular situation, then all other epistemic options she faces are morally ruled out.
  2. If the agent cannot do A, then doing A is not among her epistemic options.
  3. Hence, if an agent is morally required to do A but cannot do A, then all of her epistemic options are morally ruled out.
  4. Information that an agent is morally required to do something provides her with guidance among her epistemic options by distinguishing between options that are morally ruled out and options that are not morally ruled out.
  5. But, if all the agent’s epistemic options are morally ruled out, the moral requirement has not served its purpose by providing the agent with guidance.

It is difficult to see why premise 2 is true. As Fischer pointed out earlier, it does not follow from the fact that an agent cannot A, that she knows she cannot A. Since premise 2 cannot be motivated, the argument from the point of moral requirements should be interpreted in terms of metaphysical options. However, there are several issues with interpreting the argument this way. It seems that, if moral requirements deal with metaphysical options (options genuinely available to the agent) then there is an argument which purportedly shows that moral requirements are pointless in, and thus incompatible with, a deterministic universe. The argument is as follows:

  1. Information that an agent is morally required to do something provides her with guidance among her metaphysical options by distinguishing between options that are morally ruled out and options that are not morally ruled out.
  2. If determinism is true, then only one metaphysical option, which is entailed by the laws of nature and facts of the universe before the agent was born, obtains for any agent at any time.
  3. If an agent has only one metaphysical option at any time, then the agent cannot be guided among options, because the agent only has one option.
  4. If determinism is true, then moral requirements provide no guidance and are, therefore, pointless.

Since Copp is a compatibilist, this argument must be answered if the argument from the point of moral requirements can be successfully interpreted as dealing with metaphysical options. Copp rejects premise 2, since “agents in fact have various options – in the sense of ‘options’ relevant to moral responsibility – even if determinism is true.”9 To spell out what this sense is, we should examine Copp’s compatibilism. Copp argues: “the mere fact that a causally deterministic process leads to an upshot in which an agent does or fails to do something does not entail that the agent could not have done otherwise…even if determinism is true, not all events are determined in the same way.”10

Copp’s argument for compatibilism is based on a series of examples which involve causally deterministic processes but still invoke intuitions that lead us to believe things could have been otherwise. The first example is one in which a causally deterministic process leads to a wildfire burning through part of a forest but then stopping at a stream. Copp believes it would be coherent to say that the wildfire could have jumped the stream, since similar fires often jump even larger streams. It is commonplace to think that things could have happened differently even though they are determined to happen the way they do.

Copp then considers four examples involving a football player named Julian and a series of causally deterministic football plays. During each of the plays, Julian fails to catch a football, and each time he insists to his teammates that he could have caught it. During the first play, the quarterback under-throws Julian and the ball lands far behind him. It seems like Julian could not have caught the ball on this play. During the second play, Julian is interfered with by another player and, because of the interference, fails to catch the ball. It seems like Julian could not have caught the ball on this play, since he was interfered with by the other player. During the third play, the ball bounces out of Julian’s hands because they are too rigid. It seems that Julian could have caught the ball on this play. During the fourth play, Julian’s hands are too cold and the ball bounces out of them. It seems that Julian could have caught the ball on this play, if only he had remembered to wear gloves for the game.

An incompatibilist might try to respond that Julian was not able to catch the ball on the third play, because the ball was determined to bounce out of his rigid hands. Copp would merely respond that Julian’s hands were not paralyzed. Julian is an experienced football player, and it is intuitively plausible that he was able to catch the pass. Copp argues that this type of ability is sufficient to ground moral responsibility and allows us to preserve both PAP and the compatibility of moral responsibility in the case of determinism. This type of ability would also ground the sense of “cannot” and “option” needed for the argument from the point of moral requirements. If the term “option” is understood the same way it is an option for Julian to catch a pass that bounces out of his hands, and the term “cannot” is understood in the same way Julian cannot catch and under thrown ball, the argument could be stated as follows:

  1. If an agent is morally required to do A in a particular situation, then all other (compatible) options she faces are morally ruled out.
  2. If the agent cannot do A, then doing A is not among her (compatible) options.
  3. Hence, if an agent is morally required to do A but cannot do A, then all of her (compatible) options are morally ruled out.
  4. Information that an agent is morally required to do something provides her with guidance among her options by distinguishing between options that are morally ruled out and options that are not morally ruled out.

This new argument is only as strong as Copp’s compatibilist position, though, and I have a worry about Copp’s intuitions. It seems as if they are motivated by a distinction between what Bernard Berofsky calls type and token abilities. Julian is an experienced football player who has a type-ability to catch footballs. But at specific times, he may lack the token ability to catch footballs. He may be asleep or his hands may be too cold to catch anything. Berofsky argues that since our interest in freedom derives from an interest in moral responsibility, we should “concentrate on token rather than type ability.”11 In none of the plays Copp discusses did Julian have a token ability to catch the football, and because of this, an incompatibilist could argue it is not the case that Julian could have, in the sense relevant to moral responsibility, caught the football.

A strategy for avoiding this criticism might be to argue that type ability is relevant to moral responsibility. However, assigning moral responsibility based on type abilities seems problematic. The trolley driver whose runaway trolley runs over someone has a type ability to stop the trolley. He is a trained driver and can move his arm to pull the brake lever. However, we would not call him morally responsible for running over a person because of this type ability. He lacks a token ability because his trolley has a broken brake and is out of control. His lack of token ability is the reason we would absolve him of any moral responsibility in this case. Stating that type ability is relevant to moral responsibility does not seem to capture moral judgments like this one.

Another strategy to avoid the type/token criticism would be to spell out Copp’s compatibilist intuitions in a way that avoids using the type/token distinction. Our intuition that Julian could have caught the ball when it bounced out of his hands may not be based on his type- ability to catch the ball. Since “no two events are determined in the same way,”12 the intuition could be based on the specific nature of the causal determination that caused Julian to drop the ball. The specific causal process could be the thing eliciting our intuitive reaction. If this is the case, we would still be concerned with what happens in the individual situation, rather than what the agent has a general propensity or type ability to be able to do.

This idea of looking at the specific nature of the determinism in the causal chain is appealing because it is very similar to a position that Fischer himself holds. Fischer rejects the libertarian idea that agents need “ultimate responsibility” in order to be morally responsible for their actions. He opts instead for a view in which moral responsibility is assigned based on “indispensability.” He believes that people can have a type of control over their actions even if determinism is true. Even in a deterministic universe, people still make an indispensable explanatory contribution to their behavior. This type of contribution is, according to Fischer, enough to assign moral responsibility.13

Indispensability in the causal determination of a person’s actions could plausibly be what motivates our compatibilist intuitions in the Julian examples. When the pass is underthrown, Julian does not play an indispensable role in explaining why the pass was not caught. The explanation only needs to make reference to the quarterback who threw the ball. However, when the ball bounces out of Julian’s hands, Julian does play an indispensable role in explaining why the ball was dropped. It was dropped because Julian held his hands too stiffly. This difference in the explanatory role Julian plays might explain why we hold Julian responsible for dropping the pass in one situation but not in the other.

This indispensability approach can also adequately account for our intuitions on the blameworthiness of the trolley driver. When the brake on the trolley is broken, the trolley’s running over someone can be explained without reference to the driver: the trolley ran over the person because the brake was broken. Thus, we do not blame the trolley driver for the person being run over. But when the driver fails to pull the brake, he is indispensable in explaining the trolley’s running over someone: the trolley ran over the person because the driver failed to pull the brake. Thus, we blame the driver for running over the person.

However, remember that Fischer’s indispensability approach leads him to think that free will which involves PAP is incompatible with determinism. It is difficult to see how indispensability could rescue Copp’s compatibilism, but I will suggest a possible approach. Judgments of an agent’s ability to do otherwise (in Copp’s sense) could be relevant in determining whether an agent played an indispensable role in causing an event to occur. For example, our judgments about Julian’s ability to do otherwise could plausibly explain why he plays an indispensable causal role in not catching the ball when it bounces of his hands, but not when the ball is underthrown. Copp’s compatibilist sense of ability to do otherwise would be relevant to judging the specific nature of the causal determinism leading up to the event, and thus relevant to judgments of blameworthiness. If the ability to do otherwise can be incorporated into an indispensability theory of morality in this way, it would allow Copp to preserve the argument from the point of moral requirements, the Maxim, PAP, and his compatibilist position.

 

Bibliography

Berofsky, Bernard. “Ifs, Cans, and Free Will: the Issues.” The Oxford Handbook of Free Will. Ed. Robert Kane. New York: Oxford UP, 2002. 181-201.

Copp, David. “‘Ought’ Implies ‘Can’, Blameworthiness, and the Principle of Alternate Possibilities.” Moral Responsibility and Alternate Possibilities. Eds. David Widerker and Michael McKenna. Ashgate, 2003.

Copp, David. “‘Ought’ Implies ‘Can’ and the Derivation of the Principle of Alternate Possibilites.” Analysis 68.297 (January 2008): 67-75.

Fischer, John M. “Frankfurt-Style Compatibilism.” Free Will. Ed. Gary Watson. New York: Oxford UP, 2003. 190-211.

Fischer, John M. “Free Will and Moral Responsibility.” The Oxford Handbook ofEthical Theory. Ed. David Copp. New York: Oxford UP, 2006. 321-356.

Widerker, David. “Libertarianism and Frankfurt’s Attack on the Principle of Alternate Possibilities.” Free Will. Ed. Gary Watson. New York: Oxford UP, 2003. 177-190.

 

 

 

  1. David Copp, “‘Ought’ Implies ‘Can’ and the Derivation of the Principle of Alternate Possibilities,” Analysis 68.297 (January 2008): 69.
  2. This follows the maxim
  3. Copp, “‘Ought’ Implies ‘Can’ and the Derivation of the Principle of Alternate Possibilities,” 70.
  4. David Copp, “‘Ought’ Implies ‘Can’, Blameworthiness, and the Principle of Alternate Possibilities,” Moral Responsibility and Alternate Possibilities, eds. David Widerker and Michael McKenna (Ashgate, 2003) 274.
  5. John M. Fischer, “Free Will and Moral Responsibility,” The Oxford Handbook of Ethical Theory, ed. David Copp (New York: Oxford UP, 2006) 349.
  6. Fischer, “Free Will and Moral Responsibility,” 348.
  7. David Widerker, “Libertarianism and Frankfurt’s Attack on the Principle of Alternate Possibilities,” Free Will, ed. Gary Watson (New York: Oxford UP, 2003) 184.
  8. Fischer, “Free Will and Moral Responsibility,” 349.
  9. David Copp, e-mail message to author, September 2, 2008.
  10. Copp, “‘Ought’ Implies ‘Can’, Blameworthiness, and the Principle of Alternate Possibilities,” 293.
  11. Bernard Berofsky, “Ifs, Cans, and Free Will: the Issues,” The Oxford Handbook of Free Will, ed. Robert Kane (New York: Oxford UP, 2002) 196.
  12. Copp, “‘Ought’ Implies ‘Can’, Blameworthiness, and the Principle of Alternate Possibilities,” 294.
  13. John M. Fischer, “Frankfurt-Style Compatibilism,” Free Will, ed. Gary Watson (New York: Oxford UP, 2003) 201.

Sean Armil

Sean Armil graduated from the University of Florida in May 2009. He is currently a philosophy graduate student at the University of Arizona. [email protected]