Abstract
Brian Hedden has proposed that any successful account of options for the subjective “ought” must satisfy two constraints: first, it must ensure that we are able to carry out each of the options available to us, and second, it should guarantee that the set of options available to us supervenes on our mental states. In this paper I show that, due to the ever-present possibility of Frankfurt-style cases, these two constraints jointly entail that no agent has any options at any time. This consequence, however, is clearly unacceptable, so one of Hedden’s constraints must go. Because the ability constraint is indispensable, I argue, we have no choice but to reject the supervenience constraint. Hedden’s underlying motivation for imposing the supervenience constraint is the conviction that our options should be transparent to us, but transparency also proves to be incompatible with the ability constraint, so it must be rejected as well. I conclude by sketching an unabashedly externalist account of options, which conceives of options as exhaustive combinations of atomic movements.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In his subsequent book, Hedden (2015a: 22–28) adjusts the constraint to make it compatible with externalism about mental content. He also suggests that the supervenience base for options might extend to knowledge, in order to accommodate Williamson's (2000) thesis that knowledge is a mental state. This suggestion is puzzling: if knowledge is a mental state, my mental states supervene on nothing less than the whole of existence. Whether (for instance) my attitude that nothing can travel faster than the speed of light counts as knowledge or mere belief depends on the entire state of the universe, past, present, and future. Because supervenience is transitive, this turns (SC)* into the trivial requirement that options supervene on all of existence, which no one would think worthwhile to dispute. Hence, if (SC)* is to have any bite at all, it must be restricted to mental states which are in some sense internal.
If it seems strange that the subjective “ought” should be in the business of prediction at all, remember that standard decision theories are intended as accounts of ideal rationality, so something has gone seriously wrong if a prescription given by the subjective “ought” fails to coincide with the expected behavior of an ideally rational agent.
Cf. Conee and Feldman (2001), who argue that strong supervenience on the mental is the sine qua non of internalism about justification. A note on terminology: I will call any account of options which satisfies (SC)* “internalist,” any account which does not “externalist,” and any account of options which makes them mental entities (or propositions concerning mental entities) “mentalistic.” This is a departure from convention: typically, only internalist accounts may be described as mentalistic, but options present us with a case where mentalism is insufficient for internalism.
See also Broome (2013: 250), who proposes that we should restrict options to intentions on the basis of a supervenience argument.
We must make an exception here for cases where an agent's inability to perform some option is a foreseeable result of the agent's own prior blameworthy actions or negligence. Suppose, for instance, that Lisa causes an accident at t because she is unable to swerve in time to avoid a pedestrian. It might still be fair to blame her for the accident, if the reason she is unable to swerve at t is because she drank too much at dinner, and then chose to drive while intoxicated.
Locus classicus is Frankfurt (1969).
This resolution is not entirely satisfactory—there still seems to be something amiss about saying that an agent who fails to execute what she takes to be her best option is acting as she subjectively ought to, even if, as it turns out, she could not have done any better. In the next section, we will see that this problem is not confined to Frankfurt cases, and that it is insoluble. What we are really demanding here is that an agent's options always be transparent to her, but if our options depend on what abilities we have, this is impossible, for no non-godlike agent has transparency over her abilities.
As consolation, we might still be able to capture the intuition that the agent who passes up her best apparent option is blameworthy for acting as she does if we accept an account of blameworthiness that divorces it from the agent's abilities [Fischer and Ravizza (1998), writing on moral responsibility, suggests how this might go]. Pairing this approach with a subjective “ought” governed by the ability constraint, however, forces us to say that an agent can be subject to blame even if she executes her best option, that is, even if she behaves exactly as she ought to.
See, e.g., Lewis (1986: 86–92).
Strictly speaking, (1) and (2) need to be qualified to rule out outré cases involving duplicates who are time-travelers or who themselves have godlike powers. The needed changes are as follows:
(1)* Necessarily, for all agents and all times t, possibly there exists a non-time-traveling, non-godlike mental duplicate of that agent at t who ceases to exist immediately after t.
(2)* Necessarily, for all times t, any non-time-traveling, non-godlike mental duplicate who ceases to exist immediately after t cannot exercise any apparent options available to her at t.
(1)* will still be safely underwritten by principles of modal plenitude, and the argument remains valid, so I have suppressed this complication in the text.
The belief constraint weakens the sting of rejecting (AC), but only a little. Even if I mistakenly believe that I have the ability to cure cancer last Tuesday, it is not the case that I ought to or that I should be blamed for failing to do so. Similarly for decisions: if I will be struck by lightning a microsecond hence, it is not the case that I ought to decide to do anything or that I can rightly be criticized for failing to make the best decision, no matter what I believe. Indeed, without (AC), we are saddled with the ghoulish consequence that any agent who dies unexpectedly deserves blame for failing to posthumously execute her best option.
Proof: Suppose that M's options fail to supervene on her mental state, and φ-ing is an option for her but not for her mental duplicate, N. By the definition of mental duplication, either both M and N believe that they have φ-ing as an option, or neither does. If both do, N's belief is false, and infallibility fails. If neither does, M has φ-ing as an option without believing that she does, and self-intimation fails. In either case, options are not transparent. Therefore, non-supervenience entails non-transparency; by contraposition, transparency entails supervenience.
We must again make an exception for cases where the agent's ignorance is a foreseeable consequence of her own prior blameworthy actions or negligence. We might also wish to extend this exception to include cases where the agent's ignorance is a product of her own epistemic irrationality; I discuss this possibility further in the next section.
What of (BC), the belief constraint? If forced to choose, I believe we should replace it with a normative epistemic constraint on options, something along the following lines:
(EC): Necessarily, φ-ing is an option for an agent only if she ought to believe she can φ, given her evidence.
The chief reason for preferring (EC) to (BC) is that (EC) allows us to criticize agents who fail to execute their best option out of ignorance in cases where the agent's ignorance is caused by her own negligence in gathering or responding to evidence. I do not think we are forced to choose between the two constraints, however. We should instead follow Parfit (2011: 162–163) in decomposing the subjective “ought” into a belief-relative “ought” and an evidence-relative “ought,” where (BC) governs the former and (EC) the latter.
The dialectic is further complicated by the fact that Hedden has since disavowed the belief constraint. In a footnote (2015a: 100, fn. 9), he reveals that he could find no way around the objection that decision-theory cannot accommodate an agent's uncertainty about what her options are. This means that an agent who believes but is not certain she has the option of walking a narrow tightrope suspended across a canyon, and is in fact able to walk the tightrope, might come in for blame for failing to execute her best option, even though it seems quite reasonable for her to choose some other course of action in light of the risk. Pollock (2002) grapples with this problem.
This is true, in any case, in standard deontic logic. Some heterodox accounts of deontic modals (e.g. Cariani 2013) do not have this consequence.
Note that the ability constraint is applied twice in constructing an agent's set of options—first, to limit the atomic movements included in combinations to those under the agent's control, and second, to restrict the set of combinations which count as options for an agent to those the agent is able to perform. This redundancy is necessary because an agent may be able to perform either of a pair of atomic movements individually but not both together.
We might try to tame this excess by individuating options according to their moral characteristics, for instance, by combining all EXCAMs alike with respect to their expected utility into a single option. This proposal, though, has a fatal defect: two actions can be as different as you like, intuitively speaking, yet still share all of the same moral features. As an example, I might save five dogs from being euthanized by volunteering my nights off at the pound, or I might instead save five dogs from euthanasia by working long hours of overtime and donating the proceeds to the ASPCA, and it does not seem appropriate to conflate these two courses of action into a single option. Individuating options by their moral features will inevitably require us to run together actions which should be kept distinct.
References
Broome, J. (2013). Rationality through reasoning. Malden: Wiley-Blackwell.
Cariani, F. (2013). ‘Ought’ and resolution semantics. Nous,47(3), 534–558.
Chisholm, R. (1963). Contrary-to-duty imperatives and deontic logic. Analysis,24(3), 33–36.
Conee, E., & Feldman, R. (2001). Internalism defended. American Philosophical Quarterly,38(1), 1–18.
Fischer, J., & Ravizza, M. (1998). Responsibility and control: A theory of moral responsibility. Cambridge: Cambridge University Press.
Frankfurt, H. (1969). Alternate possibilities and moral responsibility. The Journal of Philosophy,66(23), 829–839.
Hedden, B. (2012). Options and the subjective ought. Philosophical Studies,158(2), 343–360.
Hedden, B. (2015a). Reasons without persons: Rationality, identity, and time. Oxford: Oxford University Press.
Hedden, B. (2015b). Time-slice rationality. Mind,124(494), 449–491.
Jackson, F., & Pargetter, R. (1986). Oughts, options, and actualism. Philosophical Review,95(2), 233–255.
Lewis, D. (1986). On the plurality of worlds. Malden: Blackwell.
Moss, S. (2015). Time-slice epistemology and action under uncertainty. In T. Gendler and J. Hawthorne (Eds.), Oxford studies in epistemology (Vol. 5). Oxford: Oxford University Press.
Parfit, D. (2011). On what matters. Oxford: Oxford University Press.
Pollock, J. (2002). Rational choice and action omnipotence. Philosophical Review,111(1), 1–23.
Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.
Acknowledgements
I would like to thank Peter Graham, Sophie Horowitz, and Hilary Kornblith, along with several anonymous referees, for their helpful comments on previous drafts of this paper. Special thanks to Christopher Meacham for his guidance and feedback throughout the writing process.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Koon, J. Options must be external. Philos Stud 177, 1175–1189 (2020). https://doi.org/10.1007/s11098-019-01240-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-019-01240-0