Benefits are Better than Harms: A Reply to Feit

ABSTRACT

We have argued that the counterfactual comparative account of harm and benefit (CCA) violates the plausible adequacy condition that an act that would harm an agent cannot leave her much better off than an alternative act that would benefit her. In a recent paper in this journal, however, Neil Feit objects that our argument presupposes questionable counterfactual backtracking. He also argues that CCA proponents can justifiably reject the condition by invoking so-called plural harm and benefit. In this reply, we argue that Feit’s lines of criticism are both unsuccessful.

KEYWORDS:

• harm
• benefit
• plural harm
• counterfactual comparative account of harm and benefit

1 Introduction

In a recent article in this journal [Carlson, Johansson, and Risberg 2021], we advanced an argument against the most widely endorsed and discussed theory of harm and benefit: the counterfactual comparative account (CCA). According to CCA, an event harms (benefits) someone just in case she would have been better (worse) off had it not occurred. Or, to use our more technical formulation, where ‘$W_{\mathrm{Se}}$’ refers to person S’s well-being level in the nearest possible world w in which possible event e occurs, and ‘$W_{S-e}$’ to S’s well-being level in the nearest world, relative to $w$, in which $e$ fails to occur [2021: 167]:

CCA

For any possible event $e$ and person $S$, $e$ would harm $S$ iff $W_{S-e}>W_{\mathrm{Se}}$, and $e$ would benefit $S$ iff $W_{\mathrm{Se}}>W_{S-e}$.

Our objection was that CCA violates a plausible adequacy condition on theories of harm and benefit. Informally put, the condition says that an act that would harm the agent cannot leave her much better off than an alternative act that would benefit her. More formally, where an ‘option-set’ is a set of mutually exclusive possible actions available to the agent in a given choice situation [2021: 67]:

Adequacy Condition 2

Let possible actions $a$ and $a^{\ast }$ be members of the same option-set … If $W_{\mathrm{Sa}}$ is much higher than $W_{S{a}^{\ast }}$ … , then $a$ would harm $S$ only if $a^{\ast }$ would harm $S$, and $a^{\ast }$ would benefit $S$ only if $a$ would benefit $S$.¹

To illustrate how CCA violates Adequacy Condition 2, we used the following example, which we shall here name ‘Resorts’ [2021: 168]:

Resorts

Suppose that Beth has the following four options—to visit a fantastic resort; to visit a moderately nice resort; to visit a terrible resort; or to voluntarily go to prison, where she would be stuck for life. The resulting lifetime well-being levels for Beth are ordered accordingly, with very large differences between them. As seen from the nearest possible world where Beth goes to the moderately nice resort, the nearest world where she does not do so is one where she goes to the extremely nice resort. As seen from the nearest world where she goes to the terrible resort, the nearest world where she does not do so is one where she goes to prison. (While going to the terrible resort is an option for Beth, she would not perform that action unless she had lost her preference for pleasurable experiences, in which case she might just as well go to prison.)

CCA implies that going to the moderately nice resort would harm Beth, whereas going to the terrible resort would benefit her. Yet the former action would leave her much better off than the latter. Hence, CCA violates Adequacy Condition 2. Since this condition should be accepted, we argued, CCA should be rejected.

However, Neil Feit [2022] has criticized our argument on two grounds. First, he questions our support for the claim that CCA violates Adequacy Condition 2. Second, Feit argues that by invoking so-called plural harm and benefit, CCA proponents can justifiably deny Adequacy Condition 2. We shall defend our argument against both these charges.

2 Backtracking Counterfactuals

Feit’s first line of criticism is to question the counterfactuals that we assume in Resorts to show that CCA violates Adequacy Condition 2. Consider our assumption that if Beth had gone to the terrible resort, she would have had other preferences than those which she has in the actual world (in which, we can suppose, she goes to the moderately nice resort). According to Feit, this assumption involves objectionable backtracking. While he does not spell out why, the idea is presumably either that a person’s preferences cannot realistically be supposed to change instantaneously, or that Beth’s alternative set of preferences would need to be temporally prior to her going to the terrible resort in order to explain it (as we indicated it would do). Our assumption therefore implies that if Beth had gone to the terrible resort, her preferences would have differed from her actual preferences already slightly before the time of her action. Thus, the nearest world where Beth goes to the terrible resort has a slightly different past from the actual world, relative to the time of her action. And those suspicious of backtracking will usually exclude different-past worlds from the nearest worlds, when evaluating the truth values of counterfactuals.

We agree with Feit that one should be wary of backtracking but believe that he goes too far in this respect. The very strict view of backtracking he apparently endorses implies that had Beth gone to the terrible resort, she would have acted against her preferences. In general terms, it implies that whenever we consider the possibility that an agent had chosen another option than that which she in fact chose, we should assume that her preferences would have been as in they are in the actual world. This means that most alternatives to performed actions would, if performed, have been inexplicable in terms of the agent’s preferences, as well as irrational, assuming it is a requirement of rationality to act in accordance with one’s preferences. Feit’s strict view of backtracking thus has radical and counterintuitive implications concerning rational choice and action explanation.

Those implications would perhaps be a price worth paying if a very strict view of backtracking were reasonable with regard to all other mental states and events, such as intentions and decisions. But excluding backtracking with regard to intentions and decisions has even more radical and counterintuitive implications than in the case of preferences. The assumption that if an agent had acted differently, she would still have had the same intentions and made the same decisions as in the actual world, implies that most alternative actions would have been sheer mistakes, and not really intentional actions at all. Unless we want to deny that agents often have more than one option open to them, therefore, we must allow some modest backtracking with respect to intentions and decisions.

We are not convinced, then, that the counterfactuals that we assume in Resorts are dubious. In any case, Feit also considers another example, taken from Carlson [2020: 409]:

Buttons

On a board in front of you, there are four buttons, B1 to B4, any one of which you can easily press. Pressing B1 would be very good for you and pressing B2 would be slightly less good. Pressing B3 would be very bad for you and pressing B4 would be even worse. In the nearest possible world where you press B2, it is true that if you had not done so, you would have pressed B1. Further, in the nearest possible world where you press B3, it is true that if you had not done so, you would have pressed B4.

Here, too, our objection to CCA would be that it violates Adequacy Condition 2. According to CCA, pressing B2 would harm you, whereas pressing B3 would benefit you. Yet the former option would leave you much better off than the latter. Since Feit finds the counterfactuals assumed in Buttons less problematic than those in Resorts, we shall not discuss them here.²

3 Plural Harm

Feit considers his second line of response more important. According to Feit, ‘Adequacy Condition 2 seems plausible, but it is false’ [2022: 629]. He notes that if CCA violates Adequacy Condition 2, it also violates a more general principle, which also covers an action’s effects on people other than the agent [2022: 632]:

Generalized Adequacy Condition 2

Let possible actions $a$ and $a^{\ast }$be members of the same option-set for some agent, and let $S$ be any person. If $W_{\mathrm{Sa}}$ is much higher than $W_{S{a}^{\ast }}$, then $a$ would harm $S$ only if $a^{\ast }$ would harm $S$, and $a^{\ast }$ would benefit $S$ only if $a$ would benefit $S$.

However, Feit claims that CCA proponents ‘have a good way to accommodate the appeal of Adequacy Condition 2 and the generalized version, and to replace them: that way is to adopt CCA with Plural Harm’ [2022: 632]. CCA with Plural Harm, on a characterization that Feit says is ‘incomplete but sufficient for present purposes’ [2022: 632, fn. 3], entails CCA but adds that a plurality of several events harms (benefits) someone—that is, several events together harm (benefit) her—in virtue of leaving her worse (better) off than she would have been had none of them occurred. Invoking this view, Feit says, helps to explain away the intuitive appeal of Adequacy Condition 2 and its generalized version. This is because any case in which CCA violates those conditions will contain ‘a plural harm or benefit that explains how a harm would leave someone better off than a non-harm would’ [2022: 632]. For example, Feit says, if you press B2 in Buttons, there must be something that accounts for the fact that you would have otherwise pressed B1—for instance, ‘a pro-attitude of some sort about B1’s appearance or location’ [2022: 632]. If you had neither pressed B2 nor had that pro-attitude, you would have pressed B3 or B4, and hence been worse off. Thus, Feit says, the plurality consisting of your pressing B2 and that pro-attitude benefits you on CCA with Plural Harm.

Along those lines, Feit proposes that Generalized Adequacy Condition 2 should be replaced by the following condition, which according to him CCA does not violate [2022: 633]:

Pluralized, Generalized Adequacy Condition 2

Let possible actions $a$ and $a^{\ast }$ be members of the same option-set for some agent, and let $S$ be any person. If $W_{\mathrm{Sa}}$ is much higher than $W_{S{a}^{\ast }}$, then $a$ would harm $S$ only if $a^{\ast }$ would harm S or would be among some events that harm S, and $a^{\ast }$ would benefit $S$ only if $a$ would benefit $S$ or would be among some events that benefit S.

Before we turn to the main problem with Feit’s line of defence, three initial critical remarks are in order. First, Feit’s claim that Adequacy Condition 2 is false seems inconsistent with his first line of defence, questioning whether CCA violates this condition. If Adequacy Condition 2 is false, CCA must, if true, violate it.

Second, the introduction of Generalized Adequacy Condition 2 seems dialectically pointless. While it is true that if CCA violates Adequacy Condition 2, it also violates Generalized Adequacy Condition 2 (since the latter condition entails the former), no part of Feit’s reasoning seems to rely on this fact. All his key claims can be made, and discussed, without considering cases involving more than one person.

Third, Feit provides no argument against Adequacy Condition 2. In particular, while his pluralized condition is certainly plausible, this provides no reason to abandon Adequacy Condition 2. This is because the two conditions are fully compatible; indeed, (Generalized) Adequacy Condition 2 entails Feit’s pluralized condition. Relatedly, Feit suggests, as we have seen, that the intuitive appeal of Adequacy Condition 2 can be explained away by an appeal to plural harm and benefit. But even assuming that this ‘debunking’ explanation works, this still provides no positive reason to regard Adequacy Condition 2 as false.

Still, of course, if Feit’s debunking explanation works, our argument against CCA is thereby significantly weakened. However—and this is the main problem with Feit’s line of defence—his debunking explanation is unpromising. Again, Feit suggests that the intuitive appeal of Adequacy Condition 2 can be explained away by showing that whenever CCA violates this condition, the action that would be harmful (beneficial) on CCA is included in a plurality of events that would be beneficial (harmful) on CCA with Plural Harm. To assess this suggestion properly, however, we need to make slightly more precise Feit’s rough characterization of CCA with Plural Harm (see above). As Feit himself notes in other work [Feit 2015], a plurality’s being such that a person would have been better (worse) off had no event in it occurred, cannot be sufficient for the plurality’s being harmful (beneficial). If someone misses the bus and would have otherwise been better off, she is not thereby harmed by the plurality of her missing the bus and some unrelated event that has no effect at all on her, such as a dog’s barking miles away. The version of CCA with Plural Harm developed in Feit [2015] provides the natural way to avoid such massive overgeneration of harmful and beneficial pluralities. Ignoring some complexities that are irrelevant here, this account can be formulated as follows:

CCAPH

A plurality of events $E$ harms (benefits) $S$ iff (i) $S$ would have been better (worse) off had no event in $E$ occurred, and (ii) $E$ contains no proper sub-plurality such that $S$ would have been better (worse) off had no event in it occurred.³

In the bus case, the plurality of missing the bus and the barking does not satisfy (ii).

Given CCAPH, however, it is far from clear that there are, in all cases in which CCA violates Adequacy Condition 2, any relevant harmful and beneficial pluralities of the kind that Feit’s debunking explanation invokes. Suppose, in Buttons, that you actually press B2, and that what accounts for the fact that you press B1 or B2 and would have pressed B1 if you had not pressed B2 is that you prefer using your left hand, and B1 and B2 are located to your left, whereas B3 and B4 are located to the right. Hence, the plurality of your pressing B2 and having the left-hand preference is such that if neither event had occurred, you would have been worse off. (You would have used your right hand and pressed B3 or B4.) However, this plurality is not beneficial on CCAPH. It does not satisfy (ii) since your left-hand preference is by itself such that you would have been worse off without it.

Feit therefore needs to invoke some (perhaps one-event) plurality $E$ such that you would have been worse off had neither some event in $E$ nor your pressing B2 occurred, and also such that you would not have been worse off had no event in $E$ occurred. Feit might appeal, for example, to your preference for pressing B1 rather than B3 or B4. But if this preference wholly derives from your left-hand preference, the former preference, too, is such that you would have been worse off without it. The nearest world where you lack the preference for pressing B1 rather than B3 or B4 is a world where you also lack the left-hand preference, and hence press B3 or B4.⁴ Hence, the plurality of the former preference and your pressing B2 does not satisfy condition (ii). It is highly doubtful, then, whether CCAPH implies that there is any beneficial plurality that includes your pressing B2.⁵

Furthermore, even if pressing B2 belongs to some beneficial plurality on CCAPH and pressing B3 would belong to some harmful one on CCAPH, that is still not enough to salvage Feit’s debunking response. Suppose for example that, as you press B2, you are also whistling a tune and thinking of strawberries. You would do both these latter things whichever button you were to press. However, let us add two further features to the case. First, you will be instantly killed unless you either press B2 or whistle. Second, you will be rewarded with a life in heavenly bliss if you neither press B3 nor think of strawberries (as long as you whistle). Assume, further, that if you were not to whistle you would still press B2. Moreover, in the nearest world where you press B3, you would still do so if you were not to think of strawberries. Under these assumptions, your pressing B2 and whistling is a beneficial plurality according to CCAPH, whereas your pressing B3 and thinking of strawberries would be a harmful plurality. Whether or not these implications of CCAPH are plausible, appealing to these pluralities intuitively does nothing to explain ‘how a harm would leave someone better off than a non-harm would’. At best, these pluralities seem simply irrelevant to the latter issue. But in a way, appealing to them seems even to weaken Feit’s case. If one starts out finding it implausible that pressing B2 harms you whereas pressing B3 would benefit you, although pressing B2 leaves you well off and pressing B3 would leave you badly off, it would be strange to find this more plausible upon learning that pressing B2 also belongs to a beneficial plurality (one further good thing about it) while pressing B3 would belong to a harmful plurality (one further bad thing about it).

Maybe part of Feit’s debunking strategy is that insofar as one finds Adequacy Condition 2 compelling, this is due to a failure to distinguish it from Feit’s pluralized condition, which really is compelling. It seems unlikely to us, however, for these conditions to be conflated—especially since the former is simple and elegant, and the latter is rather convoluted. For the same reason, it seems more likely that the intuitive appeal of the latter is derived from that of the former, than vice versa. In any case, the important thing is whether Adequacy Condition 2 is plausible, once clearly distinguished from Feit’s pluralized condition. And no good reason has been given to deny that it is.

Acknowledgements

We are grateful to two anonymous reviewers for this journal for their helpful comments. Thanks also to Magnus Jedenheim Edling for helpful discussion.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Erik Carlson’s and Jens Johansson’s work on this paper was supported by Grant 2018-01361 from Vetenskapsrådet and Grant P21-0462 from Riksbankens Jubileumsfond. Olle Risberg’s work on this paper was supported by Grant 2020-01955 from Vetenskapsrådet.

Notes

1 Like Feit [2022: 629, fn. 1], we have left out a clause that is irrelevant to the present discussion. We have also omitted a symbol.

2 See Carlson [2020: 409], for a brief defence of the relevant assumptions.

3 Feit [2015: 371] stipulates that a single event is always a plurality of itself. Hence, CCAPH entails CCA.

4 Similar remarks apply to Feit’s appeal to a pro-attitude about B1’s appearance or location (see above). Even if you do have such a pro-attitude, it too may wholly derive from your left-hand preference.

5 The corresponding judgment also holds for Jedenheim Edling’s [2022] variant of CCAPH. For instance, on Jedenheim Edling’s view, the plurality consisting of your pressing B2 and your left-hand preference benefits you only if for each of these events, you would not have been worse off if it, but not the other, had occurred. We can simply add to the case that if you had pressed B2 without having the left-hand preference, then you would have been worse off (for example, because you would have pressed B2 with your right hand, which would have been more uncomfortable). Jedenheim Edling’s account then yields that the relevant plurality does not benefit you. We have argued elsewhere, moreover, that there are reasons to regard Jedenheim Edling’s account as being inferior to Feit’s (Carlson, Johansson, and Risberg [forthcoming]).