INTRODUCTION
Act-consequentialist moral theories tell us to pick the best of our options – or one of the best, should it turn out that there are several optimal options. In Commonsense Consequentialism and, in more detail, in ‘Perform Your Best Option’, Douglas Portmore provides an extended defence of this idea. To be more precise, he provides a defence of a refined version of the following extremely plausible-sounding claim, which we can call ‘the maximizing principle’:
A subject, S, is permitted to perform an action, Φ, if and only if (and because) Φ-ing is an option for S and there is no other action that is an option for S that S has more reason to perform.Footnote 1
Although in the statement of this principle ‘permitted’ means ‘rationally permitted’, Portmore also defends moral rationalism – the thesis that moral requirements are a species of rational requirement.Footnote 2 As a result, the maximizing principle is of direct relevance to moral theory as well, and, if Portmore is right, it supports a version of consequentialism.
I have criticized principles such as the maximizing principle, arguing in various ways that we need to distinguish two independent dimensions of normative strength: justifying strength and requiring strength.Footnote 3 The fact that the normative capacities of any given practical reason will need to be characterized by reference to these two equally important dimensions of strength robs the simple phrase ‘more reason’ – which is crucial to the formulation of the maximizing principle and similar principles – of a univocal sense. Moreover, and as I have pointed out elsewhere, even if we read ‘more reason’ as ‘reasons with more requiring strength’, recognition of the existence of justifying strength still undermines the maximizing principle. After all, even if Φ is favoured by reasons with less requiring strength than that of the reasons that favour Ψ, it remains possible that Φ is nevertheless favoured by reasons with sufficient justifying strength to make it perfectly rationally permissible to act against the reasons to perform Ψ.
One of the virtues of recognizing a distinction between the justifying and requiring strengths of reasons is that it explains why we typically have so many rational options. That is, it explains what Joseph Raz calls ‘the basic belief’ – a belief to which Portmore is also strongly committed.Footnote 4 The explanation is the following: as long as the reasons that favour an option have sufficient justifying strength to justify acting against the reasons that favour our other options, our act is rationally justified. But the reasons that justify the action might well not have sufficient requiring strength to require it. In such a case it would also be permissible to act on the opposed reasons. As a result, the action may be rationally optional.
Portmore has recently presented his own novel and extremely interesting method of defending Raz's basic belief – in a way that is meant to be consistent with the maximizing principle. And he has also criticized one argument in favour of a substantive justifying/requiring distinction in the domain of practical rationality. It is the purpose of this article both to defend the justifying/requiring distinction and to criticize Portmore's novel maximization-friendly strategy for defending the existence of widespread rational options. If the arguments are correct, Portmore's defence of consequentialist moral theory is also undermined.
THE JUSTIFYING/REQUIRING DISTINCTION
For those unfamiliar with the distinction between the justifying and requiring strength of practical reasons, it is easy to get a quick intuitive picture by considering a similar distinction in the moral domain. One can be morally justified by considerations of self-preservation when one breaks a promise, even though breaking that promise would have been significantly immoral without that justification. That is, such a consideration has significant moral justifying strength. But a moral justification of this kind has no power to morally require anything: it has zero moral requiring strength. The opposing consideration in this case, on the other hand (that one has promised to do something), does have non-negligible moral requiring strength; absent a sufficiently strong moral justification, one would be morally required to keep one's promise. So this is a case in which one is not morally required to act on the consideration with the greatest moral requiring strength. It is also a case which illustrates the possibility of considerations that function only to justify, and shows that this need not be a result of any kind of weakness: considerations of self-preservation can often justify a very great deal.
My claim has been that the above sort of structure is also typical of the domain of practical rationality, and that many reasons can rationally justify more than they can rationally require. We can describe this view as holding that there is a substantive justifying/requiring distinction, since even Portmore would acknowledge a conceptual distinction between justifying and requiring; Portmore only denies that in the domain of rationality a reason might in fact justify more than it can require. In the present article references to a justifying/requiring distinction should always be understood to concern the substantive distinction.
The plausibility of a substantive justifying/requiring distinction can be brought out by considering two questions which one can ask regarding any given practical reason:
(R) How much can this reason require me to do?
(J) How much can this reason justify me in doing?
Even keeping in mind that the kind of requirement and justification at issue here is rational justification and requirement, and not moral justification and requirement, it is quite plausible that these questions can receive different answers. Suppose, for example, that a friend of a colleague needs to get to the airport – an hour distant – and that you have just overheard that her ride has evaporated; she needs to get on the road right away, or she will miss her flight. Suppose also that your back is hurting you, so that the two-hour round-trip would be quite unpleasant. There is certainly a reason to volunteer to drive your colleague's friend to the airport: it would save her a great deal of hassle and expense. This reason could easily justify you in volunteering to drive her, despite your back pain. But no one would claim that this reason requires you to volunteer. Nor is the existence of rational options here due to the reasons balancing out perfectly; even if your back did not hurt, you would not be rationally required to volunteer. So the altruistic reason here justifies making a non-trivial sacrifice. But that reason does not require making that sacrifice, or indeed even a significantly smaller sacrifice. So the answer to (J) in this case is ‘a non-trivial sacrifice’, while the answer to (R) is ‘not much, if anything’.
In the above case the reason that favours volunteering has sufficient justifying strength to justify suffering a certain amount of back pain, but does not have sufficient requiring strength to require it – or indeed to require very much at all. On the other hand, the reason not to volunteer does have significant requiring strength. One is rationally required not to drive back and forth to the airport – given one's back pain – unless one has quite a good reason to do so. So in this case, the reason not to volunteer seems to have more requiring strength than the reason to volunteer. And yet it is false that one ought not to volunteer. Why? Because the reason that favours volunteering has sufficient justifying strength to justify doing so. I take cases like this one to show that it is not only false or nonsensical to say that one ought to act on one's strongest reasons, but that it is also false that one ought to act on the reasons with the greatest requiring strength. The most one can say is that one ought not act against reasons with a certain amount of requiring strength unless one has reasons with sufficient justifying strength to allow it. That is, one should perform a justified option.
Portmore doubts that a reason can have more justifying strength than it has requiring strength.Footnote 5 That is, he doubts that the questions (J) and (R) above can receive different answers. In partial support of this, he cites a pair of papers criticizing my view that they can.Footnote 6 I have already addressed the arguments presented in these two papers, and I will repeat neither the criticisms nor the response here.Footnote 7 But Portmore also presents a new argument that is meant to show that while there may be an explanation for the fact that some reasons have more moral justifying strength than moral requiring strength, the same explanation is inapplicable when it comes to rational justification and requirement. Thus, one of my defences of the justifying/requiring distinction – an analogy with morality – is not, in Portmore's view, persuasive. It is worth addressing this argument, partly because it is a new one, and partly because its deflection actually provides support for the justifying/requiring distinction in the domain of practical rationality. After all, Portmore accepts that distinction in the moral domain.
Here is what Portmore says about the distinction between moral justifying and requiring strength:
It is plausible to think that agents are not always morally required to do what they have most moral requiring reason to do, because it is plausible to suppose that there are some non-moral reasons that have substantial moral justifying strength even if no moral requiring strength.Footnote 8
The non-moral reasons Portmore is thinking of here are reasons stemming from self-interest. And the sort of case he is thinking of is one in which there are morally requiring reasons to give money to charity (he holds that one would be morally required to give the money if one could do so at no personal cost) but one is justified in not giving it, because one wishes to use the money for some personal project, like brewing homemade beer. So here the self-interested reason seems to have non-trivial moral justifying strength without having any moral requiring strength. One would not be morally required to spend the money on one's project, even if there were no competing options.
Portmore thinks he has a potential explanation for the existence of a robust justifying/requiring distinction in the moral domain that fails to apply in the rational domain. And this, he thinks, allows him to accept the distinction in the moral domain, but reject it in the rational. His idea is that the considerations that provide moral justification without having any moral requiring strength are non-moral reasons: for example, self-interested reasons. So, if there were to be a similar explanation in the rational domain, we would have to countenance non-rational reasons to play that role. But the idea of non-rational reasons is (I agree) nonsense – at least if these reasons are taken to be relevant to the rational status of action. So, Portmore concludes, the existence of a justifying/requiring distinction in the domain of morality provides no support for that same distinction in the domain of rationality.
Portmore's argument obviously depends on a robust distinction between moral and non-moral reasons. Of course it is difficult to draw such a distinction in a convincing way. Indeed, it is not entirely clear whether we ought to think of moral reasons as a subset of the reasons relevant to rationality, or as a separate domain of reasons altogether: a domain that might – as some Humeans hold – have no necessary connection to rationality at all. Portmore's own view on this matter is that moral reasons are indeed a subset of reasons proper, and I won't call this view into question.Footnote 9 But the question remains: what subset? Portmore's answer is that for a reason to be a moral one, it must be able to count in favour or against an option, morally speaking.Footnote 10 Surely if a reason cannot do this, it is not a moral reason! That seems plausible enough. But the problem is how to interpret ‘count in favour, morally speaking’. If we stipulate that for a reason to count in favour of an option, morally speaking, it must have either some power to morally require or some power to make action supererogatory, then we get what Portmore wants – but only by stipulating it. And this seems to be what Portmore does. He writes: ‘A reason that could only [morally] justify – that is, a reason that could not make an act obligatory or supererogatory but could only make an act [morally] permissible – would be a (morally relevant) non-moral reason.’Footnote 11 The phrase ‘morally relevant non-moral reason’ should raise some eyebrows. Why should we not grant that if a consideration is capable of providing moral justification, then this shows that it does count, morally, in favour of an option?Footnote 12 If we do this, then Portmore's argument does not go through, since moral reasons will be capable of morally justifying without requiring, so we will have no reason to reject the same sort of claim in the domain of rationality as well.
In fact, I am not suggesting that Portmore's definition of moral reasons is inferior to a definition that includes reasons that serve only to provide moral justification. Rather, what I want to point out is that, however one decides to resolve this purely terminological issue, it remains true that some considerations can provide moral justification without having any power to make actions morally required. To say ‘But those moral justifiers come from outside the domain of moral reasons, while there are no reasons that come from outside the domain of rational reasons’ may sound convincing. But if we keep in mind that Portmore's definition of ‘moral reason’ is simply stipulative – it simply means ‘a consideration that can morally require or render supererogatory’ – we should recognize that the claim reduces to the following: ‘But those moral justifiers cannot morally require anything, while there are no reasons that cannot rationally require anything’. The second clause here makes it clear that Portmore is simply begging the question against the existence of purely justificatory reasons, and obscuring this fact with a terminological stipulation that we need not make.
In fact, even independent of its question-begging nature, Portmore's argument fails on its own terms. That is because even if we restrict our view to cases that involve only moral reasons (as Portmore defines them) we can still find instances in which we are not morally required to act on the moral reasons with the greatest moral requiring strength. All we have to do to see this is to change Portmore's own charity example, in which my justification for not acting on a moral reason came from a self-interested reason, and replace that self-interested reason with the following: that I could use the money I would otherwise give to charity to help a stranger avoid having his car towed. Surely this last reason is a moral one, even in Portmore's sense; I would be morally required to act on it if there were no reason not to. Of course it will not have as much morally requiring strength as the reason to give to charity: after all, my donation will help prevent blindness or hunger.Footnote 13 Still, I am morally justified in helping the stranger.
Let us turn now to Portmore's own ingenious method for explaining the existence of widespread rational options. After all, it could be a useful explanation even if the justifying/requiring distinction also helps explain why we have so many rational options.
PORTMORE'S PRINCIPLE PYBO
At the very beginning of this article I presented an unrefined version of Portmore's central principle; I called it ‘the maximizing principle’. In fact, the more sophisticated principle Portmore defends is the following:
- PYBO
For any subject, S, any times, t–t‴, and any maximal option for S at t, M, S is permitted at t to perform M from t′–t‴ if and only if (and because) there is no other maximal option for S at t that S has more requiring reason to perform. And, for any non-maximal option for S at t, N, S is permitted at t to perform N from t′–t″ if and only if (and because) there is some maximal option, M, that S is permitted at t to perform from t″;–t‴ such that S's performing M from t′–t‴ involves S's performing N from t′–t″ (t < t′ ≤ t″ ≤ t‴).Footnote 14
PYBO goes beyond the maximizing principle by incorporating the notion of a maximal option, and by explicitly indexing permissions to times. Moreover, because most of what we think of as our options are not maximal in Portmore's sense, PYBO also includes a principle that tells us when non-maximal options are permissible.
What is a maximal option? Here are two ways Portmore explains this notion:
Φ is a maximal option for S at t if and only if Φ-ing is an option for S at t and there is no Ψ such that Ψ-ing is an option for S at t and S's Ψ-ing involves S's Φ-ing but not vice versa. And assume that S's Ψ-ing involves S's Φ-ing just in case S's Ψ-ing entails S's Φ-ing.Footnote 15
[M]aximal options are maximally-specific sequences of actions that extend from the present to the time at which the agent ceases to be an agent any longer.Footnote 16
Essentially, maximal options are optional ways one might live out the rest of one's life. Portmore argues that we should directly assess only maximal options in terms of reasons. Then we assess non-maximal options by determining whether they are part of a permissible maximal option. This strategy can be both explained and motivated by considering a case in which a sick agent – Zeke – can cure his illness by taking one of two effective medications (A or B) during the interval from t to t‴, but not both, since the interaction of the two medications would be fatal. In such a case, the following all seem true:
(1) It is permissible for Zeke to take A at t′ (which is between t and t‴).
(2) It is permissible for Zeke to take B at t″ (which is between t and t‴).
(3) It is not permissible for Zeke to take A at t′ and B at t″.
How can we explain (3), given (1) and (2)? Portmore's answer is that we should directly assess Zeke's behaviour over the whole interval, and then assess Zeke's behaviour over sub-intervals in a derivative way, rather than the other way around. (1) and (2) are thereby entailed by the following:
(4) It is permissible for Zeke to take A at t′ and not take B at any time in the interval.
(5) It is permissible for Zeke to take B at t″ and not take A at any time in the interval.
On the other hand, the truth of (1) and (2) does not entail the following ‘agglomeration’:
(6) It is permissible for Zeke to take A at t′ and B at t″.
A generalization of this argument leads Portmore to conclude that we should assess individual actions only indirectly, by assessing the maximal options (i.e. remainders of whole lives) of which they are a part.
Portmore's ingenious explanation of the truth of Raz's ‘basic belief’ – for the existence, at any given point, of a wide range of rational options – is now easy to present. Consider one particular maximally specific way of living out the remainder of one's life that is optimal: no other way is more favoured by reasons, though there may be many that are just as well favoured. Call this way ‘M’. M is an optimal maximal option. M will contain some altruistic actions, some self-interested actions, some cookings of meals, some readings of books, and so on. But, as Portmore convincingly argues, the specific order of a set of actions in a maximal option very often does not matter to the fact that it is optimal. So there will be other optimal maximal options, M′, M″, etc., that differ from M only in that the order of some actions is switched around. As a result, at the time that the agent is deciding what to do next, there will be one optimal maximal option that includes cooking a meal now, but another optimal maximal option that includes continuing to mark papers now, and yet another optimal maximal option that includes gathering up some unwanted clothes and small appliances and taking them to Goodwill now. Because each of these optimal maximal options is permissible, so are all of the smaller options out of which they are made. I am therefore permitted, now, to cook, or continue to mark, or take things to Goodwill. This is a good result.
As we've just seen, Portmore's strategy for explaining the rational eligibility, at a given moment, of many different non-maximal options is to avoid assessing these options in terms of the specific reasons that favour and disfavour them. But surely we would also like to be able to talk about the reasons that favour taking some clothes to Goodwill, and compare those reasons to the reasons that favour simply staying in bed. Surely we would like to say that one reason for taking clothes to Goodwill is that it will help people, and that this reason does not favour staying in bed. In order to begin to address this sort of concern, Portmore offers us a way of comparing the strengths of the reasons that favour non-maximal options:
- C-PYBO
For any time, t, any subject, S, and any non-maximal option occurring subsequent to t, N, how much reason S has, at t, to perform (or to refrain from performing) N is in strict proportion to how much more (or less) reason S has, at t, to perform an optimal maximal option that involves performing N than to perform an optimal maximal option that involves refraining from performing N.Footnote 17
In order both to illustrate and support C-PYBO, Portmore points out that it implies that our sick agent Zeke, discussed above, (a) has strong reason not to take medication A at t′ and medication B at t″, but (b) has no more reason to take A at t′ than to take B at t″. The reason for (a) should be clear: an optimal maximal option that involves not taking both medications is favoured by more reason than any optimal maximal option that involves taking them both (and therefore dying prematurely). The reason for (b) is that an optimal maximal option that involves Zeke's taking A at t′ is favoured by the same amount of reason as an optimal maximal option that involves his not taking A at t′, given that such an option – being optimal – would include his taking B at t″. But, though it does capture this piece of common sense, C-PYBO has other consequences that are not as attractive. To see why this is so, it is worth reflecting on why (b) seems so plausible: both non-maximal options seem very similar, since they are both acts of taking some medication, and neither has any special benefit or harm that the other lacks. But the same logic that allows Portmore to say that they are favoured by the same amount of reason will lead to the claim that any two optional non-maximal options will be favoured by the same amount of reason. That means, for example, that Zeke's taking medication A at t′ is favoured by the same amount of reason as his staying in bed at t″, or his taking some clothes to Goodwill at t‴ instead – given that these are also optional. It is very hard to see how these claims could be true. And it is even harder to see it if we understand ‘reasons’ as we normally do: as particular considerations that favour particular actions, and that include such things as ‘that it will help someone’, ‘that it will be relaxing’ and ‘that it will cure my disease’.
And there are greater difficulties to come. Consider an action that is rationally required, but that is not particularly significant. For example, suppose it is lunchtime, one is quite hungry, but one's stomach is also slightly upset. There are two options for lunch: chicken soup or a BLT, both of which seem equally attractive in terms of taste, cost, and so on. Suppose one knows that the soup is likely to help one's stomach, and the BLT is likely to have the opposite effect. In this case, it certainly seems rationally required to choose the soup. If Portmore's account cannot yield this, it is in trouble, so let us grant that it does. What this means is that there is more reason to perform an optimal maximal option that involves eating the soup than to perform an optimal maximal option that involves refraining from eating the soup. The problem that now arises is that C-PYBO yields that there is more reason to eat the soup than there is to perform any optional non-maximal option: more reason to eat the soup, for example, than to volunteer tutoring underprivileged children after work. The reason for this is that if the volunteering is genuinely optional – which is the sort of claim Portmore is trying to capture – then an optimal maximal option that involves volunteering is favoured by the same amount of reason as an optimal maximal option that involves refraining from volunteering. In other words, it's all the same whether one volunteers after work or not, but it is not all the same whether one eats the soup for lunch or not.
CONCLUSION
Portmore might respond to the criticisms I have offered here by charging that I have ignored his argument that only by restricting direct rational assessment to maximal options can we both
(1) Avoid saying that if it is permissible for an agent to take medication A and permissible for that agent to take medication B, then it must be permissible for that agent to take both A and B, even if their interaction would be fatal.
and
(2) Embrace the idea that if a series of actions {X, Y, Z . . .} is permissible, then so must be the performance of X, the performance of Y, the performance of Z . . .
Portmore does indeed draw attention to a difficult, important and overlooked problem: how to assess individual acts and larger sets of acts in a way that doesn't lead to absurd conclusions when we put the resulting assessments together. But in my view his argument is like the epistemicist's argument that there must be some precise point at which – with the removal of a single additional hair – an initially hairy-headed person would become bald. Epistemicists certainly avoid sorites paradoxes. Despite this, however, many philosophers take the implausibility of epistemicism as a sufficient reason to reject it. I have been trying to show that Portmore's focus on maximal options yields similarly implausible implications.
None of what I have written here significantly diminishes the importance of Portmore's arguments. Just as in the case of epistemicism, a rejection of Portmore's view isn't the end of the story. Rather, it's the beginning of a search for a less problematic solution to the same problems. I do not have anything like a full solution to Portmore's challenge. But I will close by expressing my suspicion that, just as in the case of vagueness, the correct path may involve an appeal to context-dependence. In the case of practical rationality, context perhaps determines what the relevant options are. One reason for optimism about this sort of view – at least for those who think that the reasons for and against particular options are largely a matter of their consequences – is that the notions of cause and effect are themselves very plausibly context-dependent. Another reason for optimism is that the deliberative and critical contexts to which such a view would appeal typically pick out a set of relatively simple options: ones we often simply call ‘actions’. This helps explain why most theories of reasons and rationality, including my own, typically take it for granted that these are the right sort of options, and why they are typically right to do so.Footnote 18
