The real puzzle of the self-torturer: uncovering a new dimension of instrumental rationality

Abstract

The puzzle of the self-torturer raises intriguing questions concerning rationality, cyclic preferences, and resoluteness. Interestingly, what makes the case puzzling has not been clearly pinpointed. The puzzle, it seems, is that a series of rational choices foreseeably leads the self-torturer to an option that serves his preferences worse than the one with which he started. But this is a very misleading way of casting the puzzle. I pinpoint the real puzzle of the self-torturer and, in the process, reveal a neglected but crucial dimension of instrumental rationality.

Keywords:

• Cyclic preferences
• instrumental rationality
• intransitivity
• puzzle of the self-torturer
• resoluteness
• vagueness
• Warren Quinn

Acknowledgments

My thanks for helpful comments from Donald Bruckner, Matthew Frise, Preston Greene, Elijah Millgram, Michael Morreau, Doug Portmore, Theron Pummer, Andrew Reisner, Jonah Schupbach, Jacob Stegenga, Sarah Stroud, Christine Tappolet, Larry Temkin, Sergio Tenenbaum, Mariam Thalos, Ralph Wedgwood, Mike White, participants of the University of Wisconsin-Madison workshop for this special issue on Belief, Action, and Rationality over Time, participants at the 2015 work-in-progress workshop at the Philosophy Institute at the University of Saarlandes, students in my PHIL 4010 and PHIL 7500 courses, two anonymous referees, and audience members at my presentations at CRE at the University of Montreal, at the 2015 Pacific Division APA meeting, and at the 2014 Society for Applied Philosophy meeting. Thanks also to Arif Ahmed, Doug Portmore, and Sergio Tenenbaum for thought-provoking discussion (via PEA Soup) on interpreting Quinn’s position. Finally, I am grateful to the College of Humanities at the University of Utah for a travel grant supporting my presentation of the ideas in this paper.

Notes

1. See, for example, (Voorhoeve and Binmore 2006) and (Arntzenius and McCarthy 1997).

2. For a recent discussion and defense of the puzzle, see (Tenenbaum and Raffman 2012).

3. As will become apparent, the justification can hold even if, as in the case of the self-torturer, there is no threat of the agent ending up with the alternative he started with minus repeated transaction costs. Otherwise put, the justification can hold even if the scenario is such that, once an alternative is passed up, it cannot be regained for a price, and so the problem at issue is not the problem of the agent being 'money-pumped.' For the original presentation of 'the money pump argument,' see (Davidson, McKinsey, and Suppes, 1955). For my critique of the view that the money pump argument establishes that cyclic preferences are irrational, see (Andreou 2007). There, I argue that what the money pump argument shows is that an agent should not always follow his preferences regarding the options among which he must currently choose, even if these preferences are basic and the agent finds that he is stuck with them even after he is fully informed. But because the problem at issue in the case of the self-torturer is not the problem of being money-pumped, we need a different justification for why the self-torturer should not always follow his preference(s) regarding the options among which he must currently choose.

4. Quinn himself does not add the qualification 'or at least he cannot, with any confidence, determine whether he has moved up a setting just by the way he feels,' but--for reasons that I will not get into, because they are complicated and tangential given my purposes in this paper--I think the qualification is helpful.

5. If the self-torturer’s preferences are in order, then the case of the self-torturer qualifies as a 'spectrum case' supporting the intransitivity of '___ is rationally preferred to ___.' For extensive discussion of spectrum cases and intransitivity, see (Temkin 2012).

6. In 'Putting Rationality in its Place,' Quinn suggests that instrumental rationality is 'mere cleverness' and not a 'real virtue' of practical rationality if one's practical reasoning is not constrained by good ends (1993b, 234).

7. Interestingly, Quinn, at one point, maintains that 'better than... is transitive' (199). But, if 'better than' is understood as (something like) 'better in terms of serving the agent's preferences,' it is not clear that Quinn is entitled to maintain that 'better than' is transitive while also maintaining that the self-torturer's preferences are genuinely and rationally cyclic. And if 'better than' is not understood in terms of the agent's preferences, it is not clear that Quinn is entitled to assume that the relation is relevant to instrumental rationality, given his endorsement of the prevailing assumption that instrumental rationality 'is and ought to be the slave of the agent’s preferences' (209). My aim of 'uncovering a new dimension of instrumental rationality' in this paper may ultimately be of help here, but the issue is complicated and so I am working it out in a separate manuscript on the 'better than' relation (in progress).

8. The Online First version of the article, which is the version currently available, is not officially paginated, but I have added page numbers for convenience.

9. In 'Intransitive Preferences, Vagueness, and the Structure of Procrastination,' Duncan MacIntosh argues that 'if the self-torturer really has intransitive preferences… he rationally should proceed to the maximum level' (2010, 73). Relatedly, he claims that, for an agent with intransitive preferences,

each position he could have been in is such that if he does not move to a different position, he is pair-wise worse off. So, he would have been irrational to stay where he was. In moving, he has not made himself any worse off than he was before. (76)

I disagree with MacIntosh’s reasoning, but my concerns about Quinn’s take on the puzzle of the self-torturer have been influenced by MacIntosh’s thought-provoking challenges concerning the assumed irrationality of the self-torturer’s proceeding to 1000.

10. Quinn’s suggestion that it is rationally permissible for the self-torturer to end up with an alternative that serves his preferences worse than some other alternative he could have opted for is, of course, controversial. Although defending the suggestion is beyond the scope of this paper, I here accept it as plausible enough to be worth taking on board, at least for the sake of argument.

11. Keep in mind that, since that the self-torturer’s preferences are cyclic, we cannot say that his going to 1000 appears far lower in his ranking of his options (and is in this sense much less preferred) than the option of stopping at 0.

12. The discussion I am borrowing from in the next several paragraphs appears in (Andreou 2015b); there Papineau's distinction is used to illuminate the notion of parity.

13. Raffman (1994) raises this possibility and uses the distinction to argue that two color patches that are seen as belonging to different categories when judged singly can be seen as belonging to the same category when judged pairwise. This is in turn used to 'explain, in an intuitively compelling way, how a difference in kind can obtain between the endpoints (among others) of an effectively continuous series' and thus resolve the paradox in sorites cases (43). Quinn’s puzzle incorporates the assumption that, whatever the explanation, a difference in kind can obtain between the endpoints of an effectively continuous series. (More specifically, Quinn assumes that someone can go from no pain to excruciating pain via a series of unnoticeable or barely noticeable differences.) I will make the same assumption without committing to any particular explanation (though I do find Raffman’s explanation plausible).

14. Note that, although in the case of the self-torturer, the focus is on the consequences of the available alternatives, there is nothing in the idea of an appraisal response that requires that appraisal responses to potential actions be consequence-oriented; relatedly, there is, for all I say here, room for appraising an action as terrible even if it does not have terrible consequences.

15. I say a great deal more about this and consider potential objections in (Andreou 2015b).

16. For some forceful argumentation suggesting that vagueness is not crucial for supporting the possibility of rationally cyclic preferences, see (Temkin, 2012, chapter 9).

17. Relatedly, P3* figures as a plausible initial response to the worry, raised by Temkin (1996), that, given the pervasiveness of intransitivity, there may be 'no rational basis for choosing between virtually any alternatives' (209). But Temkin seems more open to the possibility of rational dilemmas than Quinn, and so Temkin may not see P3* as supporting C*, but may instead cast P3* as ensuring, in coordination with the negation of C* (and assuming that one's preferences are rational), that we are in a rational bind. As indicated above, I have accepted, at least for the sake of argument, Quinn’s view that it is rationally permissible for the self-torturer to end up with an alternative that serves his preferences worse than some other alternative he could have opted for.

18. Thanks to Sarah Stroud for pointing out that I could make do with this more modest proposal.