1. The rise and fall of satisficing

Not content with the existing forms of consequentialism, Michael Slote popularised (but later rejected) satisficing consequentialism.Footnote ² Typical act-consequentialism, which requires that one bring about the most good, seems extremely demanding. If correct, many of us are under a moral obligation to work extremely long hours, perhaps in jobs that do not fulfil us (if unfulfilling jobs pay significantly more), and donate the vast majority of our disposable income to the best causes. We could spend enough on ourselves to ensure that we are psychologically able to continue this sort of life, but our lives would be very austere.

Slote recognised that these conclusions about how demanding morality is seem to defy common sense.Footnote ³ Noting the appeal of viewing rightness as depending on the consequences of an action, he thus sought to soften the theory. No longer would the most good be required. Instead, one could permissibly just do ‘enough’. Abandoning the maximising constraint does seem like a sensible move for those sympathetic to the consequentialist project.

The moral satisficer claims that agents may, sometimes, elect to perform an action that is good enough, regardless of whether it is the best moral option. The claim is not that that a decision procedure that does not require maximising is actually best in the long run. That may be a plausible claim – efforts to the do the best might be stressful, or lead to an agent failing to achieve a good outcome due to refusing to settle for less than perfect, and ending up with nothing – but that is not the focus here.Footnote ⁴ Satisficing consequentialists accept what Tim Mulgan calls ‘blatant moral satisficing’, the claim that even ‘when the agent already knows which particular act is best from a Consequentialist perspective, she is still perfectly justified in selecting any other act which is good enough’.Footnote ⁵ It holds that it is permissible to act sub-optimally in cases where there is no moral reason that speaks in favour of doing the less good option.

Slote illustrates this with two main examples. First, he considers a hotel manager who discovers a family whose car has broken down outside the building.Footnote ⁶ They do not have money to pay for a room or a meal, but the manager offers them one of the rooms in the hotel out of beneficence. She does not offer them the best room, but simply the first suitable one she finds available, which she deems to be good enough, knowing that there are other better rooms she could find very easily.

Second, he provides an example of a fallen warrior. Because of his heroism, the gods decide to grant him a single wish for those he is leaving behind.

Slote suggests that we would not condemn the hero not wishing that they be only ‘comfortably well off’ rather than extremely well off. Even though alternative wishes would have required the same effort, Slote claims that the less generous wish is permissible.

This allows the satisficer to easily evade the demandingness objection. Agents may permissibly give some of their income to successful charities, and keep a nice amount for themselves, so long as their actions are good enough.

There are a variety of ways satisficing consequentialism can be understood. These typically vary with how the ‘good enough’ condition is interpreted. Slote actually suggests that the ‘enoughness’ will be formalisible in terms of some ‘percentage or mathematical function’ taking into account the amount of good that one could do.Footnote ⁸ This would permit agents to act in many ways we do deem praiseworthy, yet still forbid complete moral slacking off. When an agent does do more than is ‘good enough’, they act supererogatorily. There are many different ways that ‘good enough’ could be interpreted. Ben Bradley, for instance, offers six different ways.Footnote ⁹ All of the views, however, fall prey to the same devastating objection.Footnote ¹⁰

The objection is due to the permissive verdicts satisficers must give about gratuitous prevention of goodness. The typical moral satisficer must accept that it is sometimes permissible to perform an action worse than one could. In order to respond to the demandingness objection, a huge motivation (if not the entire raison d’être) for satisficing views, it must actually permit actions much worse than maximising. To see this, we can think about the type of life that the satisficer might want to think is permissible. Even if an agent gives considerably more than average to an effective charity (say 40% of a modest income), and spends three nights a week volunteering, as well as being a generally friendly, kind person in their normal interactions, they are falling way short of maximising. But surely this is precisely the type of behaviour a satisficer wants to say is permissible. So, satisficers must accept that acting significantly sub-optimally is in some (many, in fact) situations permissible.

But once we accept this, very counterintuitive verdicts follow. Tim Mulgan illustrates this with the following example:

Achilles then inputs some number (p) lower than n. Obviously, in doing so, he condemns n – p people to continued poverty. His only justification for doing so is that this number was ‘good enough’. Our response to this sort of gratuitous prevention of good is obviously unfavourable, and for most of us, this is a knock-out blow for (traditional) satisficing consequentialism.Footnote ¹²

Satisficing consequentialism is supposed to help us avoid some of the (seemingly) bad verdicts we get from maximising act consequentialism, namely that we act wrongly almost always, and that what seem like generous people actually (except when they do successfully maximise) act impermissibly. But if we accept satisficing, it looks like we must condone behaviours like Achilles's here. However, this result seems (for many of us) similarly unpalatable.Footnote ¹³

Because of this objection, the move to a satisficing view seems fruitless. As Bradley correctly concluded, to be made acceptable, satisficers must demonstrate both that their theory is well-motivated (i.e. that there is some problem it solves that extant views fail to deal with appropriately) and that they are able to provide a version that does not permit the gratuitous prevention of goodness. In the next section, I discuss the new satisficing account Chappell has formulated that appears promising.

2. Satisficing consequentialism: the reboot

Act consequentialism is often rejected because it seems to demand too much. If the demandingness is understood simply in terms of costs to welfare, the objection fails because act consequentialism actually minimises costs to welfare, once the welfare of the moral patients is taken into account.Footnote ¹⁴

Due to this type of worry, Chappell and others have speculated that we should instead understand what makes an action too demanding in terms of difficulty.Footnote ¹⁵ Chappell in particular discusses the willpower an action requires. For this reason, he describes his view as Willpower Satisficing Consequentialism.Footnote ¹⁶

Chappell holds that agents are morally required to act in a way that displays adequate moral concern. In explicating his view of moral concern, Chappell draws upon ‘Quality of Will’ accounts. These hold ‘feelings of blame to be warranted when the target of the attitude has acted in a way that displays inadequate moral concern’.Footnote ¹⁷ Specifically, Chappell understands ‘moral concern’ as a combination of two factors:

Someone with limited altruistic concern – a miserly, Scrooge-like character – thus may have to expend a great deal of mental effort to express the adequate moral concern a much more agreeable person would express easily.

Rather than outcomes, it is the moral concern an agent expresses which must be good enough. For Chappell, the level of moral concern which is good enough is settled by our theories of blameworthiness, and specifically, quality of will accounts. Greater moral concern typically promotes the general good and it is obligatory to express sufficient moral concern. More is better, but less is impermissible.

Because what moral concern counts as sufficient is settled by our theories of judgement (e.g. about the quality of will), and the moral concern a person exhibits is a function of their altruistic concern and their strength of will, Chappell is able to express the maximum effort an agent must expend in a situation. He calls this the effort ceiling, which he defines as follows:

To elucidate, we have some prior notions about how much people need to do with regard to promoting the general good. Moral concern is seen as a combination of both altruistic concern and strength of will. A particularly selfish agent will have minimal altruistic concern, so to meet X, they may need to put in a great deal of willpower, whereas someone who is particularly other-regarding (having a high amount of altruistic concern) may need a very low amount of willpower to meet X. It does not take much willpower on behalf of an altruistically disposed agent in order for them to act well, whereas the miser may find it extremely motivationally difficult.

Chappell makes use of this effort ceiling in order to offer his version of willpower satisficing consequentialism.

Any act which causes less utility than could be produced with X, or which does not produce as much utility as could be provided for the amount of effort expended, is thus regarded as impermissible. The limits of moral demands are set by the willpower effort ceiling. With this in mind, we can see that a prescription is overly demanding if it requires more willpower than X.

As Chappell notes, this gives felicitous verdicts in the cases of gratuitous prevention of goodness. The standard moral satisficer must accept that it is sometimes permissible to perform an action much worse than one could, at a whim or for no reason, and even though it is just as easy – requires just as much exertion of the will – as other options. Chappell is able to avoid this unpleasant consequence. Consider Mulgan's Magic Game example. The effort ceiling, X, in this case, is the (minimal) effort of Achilles inputting the numbers. WSC then requires that he do as much good as possible, with this amount of effort (or perform some morally efficient act which does more good). As we would hope, WSC will give us the verdict that Achilles must input the number n.

Mulgan actually suggested that an ‘effort satisficing’ view – like Chappell's – was possible in The Demands of Consequentialism, noting a temptation to ‘think that Slote can construct a Satisficing Consequentialism based on the notion of “good-enough effort”’.Footnote ²⁰ However, this is definitely not what Slote had in mind. In fact, Chappell's account could not give the verdicts that Slote favours in several of his examples. This might not be too much of a problem for WSC, as these cases (the Hotel case and Fallen Hero) are not the sort that motivate satisficing consequentialism (i.e. instances where the demands placed on the agent by act consequentialism seem unreasonable).

Mulgan raises a potentially bigger worry, that Slote could not make the ‘good enough’ concern effort rather than outcomes, because it would not provide a consequentialist explanation for the permissibility of sub-optimal acts:

However, if the reason we are permitted to act sub-optimally is because it is accepted that we are not required to expend unreasonable efforts, this is no longer justified, at its core, in terms of consequences. A major motivation for satisficing consequentialism is the general appeal of consequentialism but a dislike of the maximising aspects, so if the resulting theory is no longer consequentialist – explaining obligations in terms of values of outcomes – this might be seen as a significant cost for WSC.

I don't think this is a major concern. The WSC formulation does define rightness in terms of utility, so it does seem appropriate to call this consequentialist.Footnote ²² My worry, however, is not that the theory isn't consequentialist enough, but that it still fails to give the right verdicts. In the next section I suggest that Chappell's account incorrectly assesses as impermissible actions which are good, but not as good as others which require the same effort.

3. Responding to Chappell

A feature of Chappell's account is that it requires that any supererogation be efficient. While this is accurate in some cases, I argue that it overgeneralises. Sometimes, an agent can do more than required without having to do as much good as possible for the willpower exerted.

Of course, we can grant that acting in a way to deliberately and gratuitously cause less good than one might is often a bad thing. This is clear in Mulgan's Achilles case. We can also notice this in cases where an agent seems to be under no obligation to act at all. Consider Kagan's case where an agent has to choose whether or not to run into a burning building to rescue survivors.Footnote ²³ We may accept, given the level of risk that they would face, that it is perfectly permissible for an agent not to enter, i.e. to do nothing (expending zero effort). However, if they do choose to enter, and once inside they have a choice to save either a bird or a baby, it seems that they are under an obligation to save the baby, rather than the bird. Doing nothing is fine, running in to save the baby is fine, but running in to save the bird (we typically think) is not okay. This is an instance of gratuitous sub-optimal supererogation.

Chappell is also able to get the correct verdict here. If one is under no obligation, the X value would be zero, but if one expends the effort to run into the building, one must produce no less utility than any alternative one might perform at the cost of up to that willpower. Again, WSC looks good. However, while WSC successfully condemns gratuitous prevention of goodness, it also condemns any supererogation which is inefficient, i.e. that does not produce the most possible utility for the effort expended. This causes some problems, as not all instances of inefficient supererogation seem morally pernicious.

Consider philanthropic giving. While squandering one's money by donating to a charity one believes foolishly wastes its donations (but does some good) strikes us as clearly bad, many of us would recoil at the prospect of condemning anyone who gives a substantial amount of their income to a not-perfectly-efficient charity – perhaps even the second-best charity. Chappell's account requires that any time we go above the effort ceiling, we are morally required to do the best we could at that effort level. This seems implausible. If someone did give all their disposable income to the second best charity, it would seem extremely inappropriate to condemn them for doing so (discarding any other bad-making moral features).Footnote ²⁴

We can even consider much more mundane examples. Imagine the minimum you can do for Mother's Day is send a card and make a phone call. You realise also that the best thing to do would be to take her out to her favourite restaurant (Mother's Choice). However, a close second would be to take her out to a restaurant you prefer (Child's Choice), which she also likes. The amount you prefer the Child's Choice (by stipulation) is slightly less than the amount your mother prefers Mother's Choice. Assuming attending either restaurant would take the same amount of effort – the price of the food is similar, the distance to travel is the same, etc. – WSC is committed to condemning taking her to Child's Choice. But you are permitted to simply call and send a card, which generates much less utility!

In this case, there is something that counts in favour of performing the act with less utility. You prefer the food at this restaurant. We might think that your knowing this could make it the case that going to Mother's Choice is slightly more difficult – motivating oneself to go there takes a little bit more of your willpower. Unfortunately, if an agent's mild preferences are taken to affect willpower in this way, WSC loses its ability to give the felicitous verdicts in many of the previous problem cases for satisficing consequentialism. Take the bird and the baby case. If an agent has a particular fondness of birds which would make it very slightly more unpleasant for them to leave the bird, they would be permitted, according to WSC, to opt to save the bird over the baby. This is a very undesirable verdict.

Chappell could respond to examples like these by amending the second clause of WSC, which says that for actions requiring effort more than the effort ceiling, the act must promote as much utility as possible at that willpower cost. He might, for instance, remove it:

However, doing so would lead to a big problem, as this would condone all sub-optimal supererogation, even where this looks dubious, as with Kagan's bird-saver. Another example where this looks like the wrong result is in Joe Horton's illustration of the ‘all or nothing problem’, where an agent has the following options:Footnote ²⁵

1. 1. Do nothing.

2. 2. Save one child by allowing both your arms to be crushed.

3. 3. Save two children by allowing both your arms to be crushed.

Because of the huge cost involved, it seems permissible to do nothing here, so for WSC no effort is required. If WSC was amended to require only actions that produce the maximum utility at that level of effort (none), any of the other two actions would be permissible. However, Horton contends, plausibly, that it is obviously wrong to decide to save only one child, when one could – at no extra cost or effort – have saved two.

Another way Chappell might amend WSC would be as follows:

Where C stipulates a certain relationship between, for instance, the additional utility gained for an agent and the utility gained external to the agent. This could function like Scheffler's agent-centred prerogatives,Footnote ²⁶ so that above the effort ceiling an agent could count their own interests as more weighty than others’ interests. This could even be supplemented with constraints on harming others to avoid verdicts such as it being permissible to expend effort above the effort ceiling to harm someone else by a similar amount it would benefit an agent by.

However, whatever way such a condition, C, is construed, if it only operates over values of personal and impersonal utility produced, it still seems unsatisfactory. To see why, consider the following two cases.

1. 1. Take Kagan's example of the bird and the baby; however, in this case the bird is mine. Because of the danger it is permissible to do nothing. It is also permissible for me to run in to save the baby. It is not permissible to save the bird. Even if the bird is mine, saving the bird seems impermissible. Thus, the condition should not deem it permissible to make myself slightly happier to expend some effort in saving a bird rather than a baby.

2. 2. I donate a significant sum of money (supererogatorily) to the second-best charity. I opt for this because I know someone who suffers from a disease it treats, and this gives me a warm feeling inside. We typically view this as permissible.Footnote ²⁷

Let me stipulate that the warm feeling in case 2 is, in terms of my personal utility, the same as I would gain from saving the bird in the first case. The difference in the alternative costs, however, is significantly higher in the charity example. The second-best charity might be able to save hundreds of lives fewer with the money I donate. So any condition simply operating over utility values could not allow WSC to give the desired verdicts.

Perhaps a more promising way of amending WSC could be to replace the second clause of WSC with something like an ‘Avoid Gratuitous Worseness’ principle.Footnote ²⁸ This could impose a requirement that any inefficient giving is not simply gratuitous, i.e. that something counts in its favour. If this was not sufficiently strong, however, this would permit one to save the bird rather than the baby, or even save only one child's life in Horton's case, so long as the agent can find some reason that counts in its favour. Developing an account of what makes certain acts objectionably gratuitous would then be a challenge. However, a satisfactory answer to this question looks set to render the rest of the WSC apparatus redundant, as this would seemingly tell us everything that WSC sought to explain.

Another response available is accepting that WSC clashes with some common intuitions, but still finding it suitable. We could suggest that intuitions are misleading in certain situations, so it is reasonable to expect some verdicts that clash with our judgements. In the philanthropy case, effective altruists and others might welcome the verdict that giving sub-optimally actually is morally impermissible.

However, I think we have independent reason to reject WSC, namely that it can condone certain actions justified by self-interested reasons, but is unable to permit better actions, which would carry similar justifications. To illustrate, consider again the Mother's Day case. Presumably, it is because I am allowed to give some weight to my own self-interested reasons that I am morally permitted to send merely a card and make a phone call. So self-interest permits me to exert less willpower, and consequently to produce less good. WSC can accept this, as our blaming practices permit some self-interested behaviours. But, the same explanation – my self-interest – does not permit doing less good when I do want to put more effort in. Self-interest, under WSC, can therefore permit doing no good at all (or negligible good, in this situation), but not doing significant good inefficiently.

For WSC then, self-interest can justify omitting to act, but not justify choosing one option over another at the same willpower. It would seem bizarre if the permissibility of my acting slightly self-interestedly depends only on how much effort I exert.

4. Concluding remarks

Chappell's willpower satisficing successfully avoids condoning the gratuitous prevention of good. This was a disastrous and unacceptable result. The examples I have provided show that WSC also gives strange verdicts. Denying the intuitive verdicts in the cases I have supplied, as required by WSC, would not require as revisionary treatment of our moral landscape.

But I do regard the verdicts as a problem, and a serious one. Chappell has, by implanting the ‘quality of will’ and willpower concepts, resurrected satisficing consequentialism into a stronger candidate for a moral theory. However, the account still fails to match up with huge classes of typical moral intuitions.Footnote ²⁹