4.1 Preamble

There are various ways in which a defender of the Patterned View might respond to the argument above. I examine what I take to be their strongest options in a moment. However, before doing so it is important to underline one limit to the kind of counter that such a proponent can offer. As emphasized above, the principal claim that this paper is looking to assess is the claim that one advantage the Patterned View enjoys over numerous other pattern-dependent views – most notably, traditional rule consequentialism – is that, unlike those views, the Patterned View avoids the Ideal Worlds objection. This being so, when assessing potential ways a proponent of the Patterned View might respond to the argument above, we need to make sure we do not allow them to make any moves that are also available to advocates of other pattern-dependent views. Otherwise, those advocates might reasonably claim the Patterned View offers no advantage over their view. In a slogan, we can say

Let us now consider the possible responses a proponent of the Patterned View might offer.

4.2 ‘Going general’

One way that proponents of the Patterned View might respond to the counterexample offered above is to claim that the best version of the Patterned View is unlikely to recommend rules as specific as (B). Rather, given the Generality Requirement, it would likely recommend much more general rules. Indeed, in his paper, Perl notes one rule he could appeal to here.

Assuming the Patterned View recommends the Benefit Rule, it will not recommend that I press Green in Reluctant*. Pressing Green does not choose the greater benefit over the lesser. Rather, I should press Red.

For a proponent of the Patterned View to make this move is, of course, for them to endorse the wide reading of the Generality Requirement noted earlier – that is, one which advantages those rules that opt for more inclusive description of the relevant act. I think this move may cause problems for them elsewhere. However, there are a couple of more immediate problems with this solution.

First, ‘going general’ is a solution available to other pattern-dependent theorists. Both traditional rule consequentialism and, arguably, Scanlonian contractualism have principled reasons for preferring more general rules/principles to more specific ones. Moreover, such theories can be understood to recommend broad duties of beneficence in the style of the Benefit Rule – especially when, as Perl stipulates with regards to the Benefits Rule, such duties are ‘subordinated’ to other rules (Perl Reference Perl2021: 122). ‘Going general’, therefore, violates No Specious Arguments.

Second, ‘going general’ does not solve the problem. Rather, in response we might simply change the case. Consider

However, suppose also that, just before we begin our deliberations, you explain that you're not going to adjudicate properly, because you do not want to.

Here, again, the Patterned View is open to the Ideal Worlds objection. Presumably here a proponent of the Patterned View will want it to recommend ‘Adjudicate Properly’ (Perl's chief example of a rule defended by the Patterned View is a very similar ‘Fair Trials’ rule). Yet where I adjudicate properly and you do not, my act would make things go very badly, and for no good reason.Footnote ⁶

4.3 ‘Biting the bullet’

At this stage, perhaps, a proponent of the Patterned View might simply try to bite the bullet. Here they might remind us of the fact that the Patterned View is a principled view. It is not Act Consequentialism. As such, there will necessarily be cases in which it recommends acts that do not make things go best. Nevertheless, such a proponent might argue, it correctly identifies our moral duty. For example, in Judges I ought to adjudicate properly, even where you do not.

However, ‘biting the bullet’ is also a move that many other pattern-dependent theorists will want to make in precisely the kinds of cases under discussion. So, it violates No Specious Arguments.

4.4 ‘Going conditional’

An alternative way proponents of the Patterned View might look to avoid the counterexamples above might be by employing conditional rules. (Some things Perl says suggest that this is the kind of strategy he thinks proponents of the Patterned View ought to employ.)

We can see how this might work. Take Reluctant*. In Reluctant*, a proponent of the Patterned View might claim that we do not have to see the relevant choice as a narrow one between

1. (A) Press Red, not Green;

and

1. (B) Press Green, not Red.

Rather, we might see it as a choice between (A), (B) and

1. (C) Press Green, unless some people are pressing Red, in which case, press Red.

From here they might argue that the Patterned View will not only prefer (B) to (A) but also (C) to (B). First, the Patterned View will credit (C) with the same good it credits (B): the lives saved through successful coordination in pressing Green. However, it will also credit (B) with some bad it does not credit to (C): the lives lost through my dogmatically continuing to press Green when others are pressing Red. So, since the Patterned View will credit the same good to both (B) and (C) but some bad to (B) that it does not to (C), it prefers (C) to (B). And where the Patterned View prefers (C) to (B), its proponents can claim that it is not open to the Ideal Worlds objection: it does not require that I act in certain ways even when, because you are not acting in the same way, my act would make things go very badly, and for no good reason.

Despite this apparent success, however, it seems clear that ‘going conditional’ will not help the Patterned View avoid the Ideal Worlds objection. As traditional rule consequentialists have found, pattern-dependent views of the relevant sort face a series of problems when looking to resolve the Ideal Worlds objection by employing conditional rules. At least some of these problems also affect the Patterned View. Given the nature of the problems in this latter set, we have reason to believe that the Patterned View does not enjoy a significant advantage over traditional forms of rule consequentialism, with regard to avoiding the Ideal Worlds objection.

For example, one problem writers have noted when traditional rule consequentialists ‘go conditional’ is that they can struggle to ensure that their theories will actually adopt the kind of rules that would help them in the relevant range of cases. By tweaking our counterexamples, we can ensure that the consequences of distant patterns of action (patterns that, here and now, we do not have the power to bring about through our own action, nor have the power to place a burden on others to bring about) will push such theories to endorse rules that their proponents would not want them to endorse (say, rules like (B) in Reluctant*) rather than rules that they would (like (C)).

To see this, take URC and a case like Reluctant*. Now, suppose that we stipulate that there will be fantastic consequences if everyone obeys (B). (To help us envisage such a scenario, Podgorski (Reference Podgorski2018) asks us to imagine something called a ‘utility landmine’ – that is, a device that is completely inert until a particular trigger condition is met, where upon, if it is a ‘goodmine’, it will create a near infinite quantity of good things; and if it is a ‘badmine’, it will create a near infinite quantity of bad things. Here, we are imagining there is a goodmine triggered to go off if everyone obeys (B)). In this case, despite proponents of URC hoping their theory might endorse conditional rules like (C), it will prefer (B): in the world in which URC tests its rules – that is, a world of universal adherence – (B) has better consequences than (C). Yet, that brings us back to the Ideal Worlds objection. Once again, URC would wrongly require me to act in certain ways in Reluctant* (i.e. dogmatically follow (B)), even when, because you are not doing the same, my act would make things go very badly, and for no good reason. Call this the Misdirection Problem.

The Misdirection Problem is a significant problem for traditional forms of rule consequentialism when they try to solve the Ideal Worlds objection by ‘going conditional’. However, with respect to that problem, the Patterned View seems to have much the same weaknesses. For example, suppose that, years after we have played our game in Reluctant*, I invent a badmine that triggers whenever someone obeys (C) and then that, later that week, someone does obey (C) and the badmine is triggered. In that case, the Patterned View would prefer (B) to (C): in the world in which the Patterned View tests its rules – that is, the past, present and future of the actual world – (B) has better consequences than (C). Yet, that brings us back to the Ideal Worlds objection. Once again, in the case of Reluctant*, the Patterned View would wrongly require me to act in certain ways (i.e. dogmatically follow (B)), even when, because you are not acting in the same way, my act would make things go very badly, and for no good reason.

It may be that a proponent of the Patterned View wants to take issue with the last bit of that sentence – that is the bit that says the Patterned View is directing us to undertake certain actions ‘for no good reason’. We will get to that point in Section 4.5. For the moment, however, we need only register that the structure of the problem, and corresponding objection, is the same. In both cases, the candidate moral theory is directing us to acts which, because others are not acting in the same way (following (B)), will generate harm (saving none where we could have saved some).

In one respect, I do not think we should find this result – namely, that the Patterned View is as susceptible to the Misdirection Problem (and, with it, the Ideal Worlds objection) as traditional forms of rule consequentialism – terribly surprising. This is because, as far as I can see, both the Patterned View and traditional forms of rule consequentialism share the same structural weakness. This is a key respect in which my analysis diverges from Perl's. According to Perl's diagnosis, the reason traditional forms of rule consequentialism are vulnerable to the Ideal Worlds objection is because they test and validate their codes by examining the consequences of hypothetical patterns of action (hence, presumably, his desire to eliminate such considerations). If this were true, then the Patterned View could avoid the Ideal Worlds objection simply by eliminating all talk of hypotheticals. However, as the analysis above shows, the problem is clearly wider. As indicated in Section 2, what gives the Ideal Worlds objection bite against traditional forms of rule consequentialism is not that they test and validate their codes by examining the consequences of hypothetical patterns of action. Rather, as uncovered by writers like Podgorski and others, it is that they test and validate their codes by examining the consequences of distant patterns of action. That is, patterns of action that, here and now, we do not have the power to bring about through our own action, nor place a burden on others to bring about. (This, I take it, is one of the reasons Podgorski strongly encourages us to re-title the Ideal Worlds objection as the Distant Worlds objection – or, as I might prefer, the Distant Patterns objection). It is this feature of such theories that makes, in Podgorski's phrase, traditional forms of rule consequentialism liable to ‘miss the landmines at their feet’. Yet, it is also in this respect that there is no difference between traditional forms of rule consequentialism and the Patterned View. Thus, both are vulnerable to the same objections.Footnote ⁷

Before moving on, let me briefly deal with another point. Earlier I mentioned that there are certain problems traditional forms of rule consequentialism face when ‘going conditional’ that the Patterned View avoids. Here is one example. Work by Smith (Reference Smith2010) shows that one problem traditional forms of rule consequentialism face when looking to test codes containing conditional rules is that they struggle to assign them a determinate value. Smith's examples are perhaps too intricate to rehearse here. However, in brief, the problem arises from the fact that there are multiple ways in which a group of agents may adhere to a code made up of conditional rules and each of these patterns of action can have very different consequences. This might not be an issue if traditional forms of rule consequentialism told us which patterns of action obtain in the worlds in which they test their rules. But they do not. They only tell us that, when they test rules, they do so in worlds of such-and-such adherence. On traditional rule consequentialism, therefore, the expected value of codes containing conditional rules – the extent to which they tend to result in good or bad consequences – remains indeterminate. We cannot tell whether one code made up of multiple conditional rules is better than another. Call this the Indeterminacy Problem.

The Patterned View avoids this problem. Unlike traditional forms of rule consequentialism, the Patterned View does tell us which patterns of action obtain in the world in which it tests its rules: namely, that pattern of action that does obtain (or did, or will, or whatever). Thus, on the Patterned View, the value of some code will always be determinate (provided that what obtains in the actual world is determinate), even if the code being tested contains multiple conditional rules. There are, then, at least some problems traditional forms of rule consequentialism face when ‘going conditional’ that the Patterned View avoids.

Does this, then, vindicate Perl's claim that the Patterned View offers a near-Pareto improvement over traditional forms of rule consequentialism? Yes and no. The crucial question here is what we take that claim to imply and, indeed, how ambitious proponents of the Patterned View are about the prospects of their theory.

First, there is a limited sense in which, on the grounds set out above, proponents of the Patterned View can claim the View offers a near-Pareto improvement over other forms of rule consequentialism. Namely, they might claim that it is ‘better on some questions – e.g. the Indeterminacy Problem – but worse on none’.

Now, as I shall show in Section 4.5, I think there are reasons why we might doubt even this claim. (I would argue that the Patterned View is better on some questions but also that it is worse on some questions.) However, leaving this point aside, as indicated in the Introduction, I take Perl's claim to be more ambitious than this limited reading of what might constitute a ‘near-Pareto improvement’. That is, I take his claim to be not simply that the Patterned View is ‘better on some questions and worse on none’ but that the Patterned View has an advantage over traditional forms of rule consequentialism in that, unlike those positions, it avoids the Ideal Worlds objection. That is a significant claim. Moreover, it is significant not only because it is one problem faced by certain kinds of rule consequentialists (as is the case with the Indeterminacy Problem). Rather, it is significant because many philosophers consider the Ideal Worlds objection to be a fatal objection to the rule consequentialist's entire project. Were this claim to be vindicated, therefore, it would be of interest not only to rule consequentialists but also to non-rule consequentialists. It would give them reason to sit up and take notice. Yet, as the analysis above shows, even when it ‘goes conditional’, the Patterned View does not avoid the Ideal Worlds objection. It still sometimes requires us to act in ways that go badly, and for no good reason. Thus, I take it that, with respect to that latter, more ambitious and interesting claim, the Patterned View does not offer us the Pareto-improvement we were looking for.

4.5 Rewriting the Ideal Worlds objection

Let us consider one last way a proponent of the Patterned View might look to respond to the arguments above. One thing they might argue is that it was never their intention to claim that the Patterned View avoids the Ideal Worlds objection in the terms in which we have been discussing it. Indeed, they might even concede that, so described, the Patterned View fares no better with regards to that objection than traditional rule consequentialism. Yet, they might claim that there is another, narrower version of the objection with respect to which the Patterned View does enjoy such an advantage: namely, one specifically addressed to those pattern-dependent theories that ground our reasons for action in facts about hypothetical worlds. Say,

An example can help us make sense of this move. In his (Reference Podgorski2018), one of the ways Podgorski helps us to see the force of the Ideal Worlds objection is through the idea of a dud. That is,

Let us imagine there are some set of duds the trigger conditions of which are restricted to hypothetical worlds/patterns of behaviour, and another set the trigger conditions of which are restricted to actual worlds/patterns of behaviour. Call the former h-duds and the latter a-duds. From here, a proponent of the Patterned View might claim that, even if the Patterned View is susceptible to a-dud counterexamples (along the lines of, say, Reluctant*), it avoids h-dud counterexamples (along the lines of, say, the original Reluctant).

There are, however, several problems with this move. First and foremost, even if it were successful, such a move would leave uncontested the main thesis of this paper: namely that the Patterned View remains open to a standard interpretation of the Ideal Worlds objection. That would be good news for this paper. But it would be bad news for proponents of the Patterned View. As described in Section 4.4, many philosophers view susceptibility to any dud counterexamples – h-dud or a-dud – to be fatal to a theory's prospects. After all, it entails that the candidate theory will sometimes require us to make things go badly, and for no good reason. By failing to challenge the idea that the Patterned View is so susceptible, therefore, the proponent of the Patterned View would effectively undercut any motivation any non-rule consequentialists might have for taking an interest in the theory. By their way of thinking, the dispute between the Patterned View and traditional forms of rule consequentialism would be best understood as simply a disagreement between similarly doomed positions.

Second, even if we understand the proponent of the Patterned View's argument to be simply an argument about which out of (similarly doomed) forms of rule consequentialism is best, it is still not entirely clear why we should think the mere fact that the Patterned View is not susceptible to h-dud counterexamples suggests it has an advantage over more traditional forms of rule consequentialism. After all, in response, traditional rule consequentialists might make a similar claim. They might argue that their position offers an advantage over the Patterned View in that it avoids a-dud counterexamples.

In response, a proponent of the Patterned View might attempt to argue something like:

Some things that Perl says in both his (Reference Perl2021) and elsewhere suggest that this is the kind of argument he wants to make. Yet I am not sure his reasoning here is entirely convincing.

For example, one thing Perl says in his (Reference Perl2021) is that the problem with the way traditional rule consequentialism grounds moral facts in facts about hypothetical patterns of action is not only that it means such theories will sometimes direct us to undertake harmful actions but that, when they do, they make such recommendations on the basis of facts that are, in some way, ‘irrelevant’, or something ‘it makes no sense’ for us ‘to care about’.Footnote ⁸

One way we can understand this argument is as leaning into the latter part of the Ideal Worlds objection. That is, where that objection argues first, that

and second that

Perl seems to want to maintain that, irrespective of the susceptibility of either the Patterned View or traditional rule consequentialism to the first part of the objection, traditional rule consequentialism is more liable to the second part than is the Patterned View. The thought is that, at least when the Patterned View makes its demands, it does so on the basis of ‘relevant’ considerations, or something ‘it makes sense for us to care about’.

Yet, the operative sense of ‘relevance’ here is obscure. Indeed, on reflection, we might think there is a strong sense in which both views make their demands on the basis of something ‘irrelevant’. (So the Ideal Worlds objection is vindicated to that extent.) Namely: both views claim we ought to follow certain rules on the grounds that, in certain circumstances, adherence to such-and-such a rule realizes certain consequences, and that is true even though, here and now, there is no chance of those consequences being realized.

For example, take URC. In Section 4.3, we saw that, in cases like Reluctant*, URC can demand that I follow (B) over (C) because, in the places where URC tests its rules, (B) has better consequences (it sets off a goodmine). Yet, this is even though there is no chance of such consequences being realized by my following (B) in that case (after all, in Reluctant*, not everyone is following (B)).

The Patterned View has the same problem. In Reluctant*, the Patterned View can demand that I follow (B) over (C) because, in the places where the Patterned View tests its rules, (B) has better consequences (it doesn't set off a badmine). Yet, this is even though there is no chance of such consequences being realized by my following (B) in that case (after all, in Reluctant* I haven't even invented the badmine yet). In this way, then, we might think that irrelevance is always going to be a problem for both theories.

Say a proponent of the Patterned View were to concede this. Still, they might argue that there is still a difference between the two cases (one that vindicates the Asymmetry thesis), in that the Patterned View draws its instructions from patterns of action that will one day happen (I am, one day, going to set off that badmine). By contrast, traditional rule consequentialism draws its instructions from patterns of action that may never happen (it may be that the goodmine is never set off).

Yet, again, it is not clear why this fact should mean that what grounds rules on the Patterned View is something ‘it makes sense for us to care about’ where the same is not true of traditional rule consequentialism. Note, here the proponent of the Patterned View cannot simply argue that, if some rule, R, is to demand an act for good reason, it must be the case that the public goods that ultimately justify R will one day be actual. To assert such a claim is to beg the question against the traditional rule consequentialist.

Elsewhere in his writings, Perl suggests there may be a way in which proponents of the Patterned View can defend this kind of claim without begging the question. That is, he argues that even traditional rule consequentialists should agree that, if some rule, R, is to demand an act for good reason, it must be the case that the public goods that ultimately justify R will one day be actual, because that follows from their commitment to consequentialism. By virtue of being consequentialists, Perl argues, traditional rule consequentialists should agree that ‘the moral facts supervene on the consequences of agents’ actions' (Perl Reference Perl2023: 907). Yet if they agree to this, he goes on to intimate, they should also agree that if some rule, R, is to demand an act for good reason, it must be the case that the public goods that ultimately justify R will one day be actual.

However, there is a problem here. We might agree with Perl that any consequentialist must agree that the moral facts supervene on the consequences of agents' actions. However, we might also think that consequentialists can disagree amongst themselves about whether the moral facts need to supervene on the consequences of agents' actual actions. Some consequentialists might hold that the moral facts supervene on the consequences of agents' hypothetical actions. Such a position is recognizably consequentialist (it is talking about consequences of agents' actions). Yet it does not entail that if some rule, R, is to demand an act for good reason, it must be the case that the public goods that ultimately justify R will one day be actual.Footnote ⁹

Here I acknowledge that there may be some readers who agree with all of this and yet who might still think the proponent of the Patterned View has a point. That is, they might think there is something odd about the way traditional rule consequentialists ground their rules in patterns of action that may never happen. Yet, in response, traditional rule consequentialists can make an analogous argument. They can argue that there is also something odd about the way the Patterned View grounds morality on only those patterns of action that actually happen.

To see this, consider extramarital affairs. Now, I cannot claim to have any special, first-hand knowledge of extramarital affairs (I promise!). However, it seems plausible to suppose that the overwhelming majority of the agent-neutral bad generated by an extramarital affair is generated upon its discovery. Perhaps there is some bad generated while the affair remains secret. Maybe the unfaithful parties have feelings of remorse and shame. Perhaps the person cheated on has a sense there is something amiss, or is prevented from enjoying certain goods that they might otherwise enjoy (care and attention and so on). But it seems likely that the majority of the hurt and pain associated with such affairs is caused only once the affair is uncovered. If that is correct, and if we also suppose that while they remain secret, such affairs generate a net gain of agent-neutral good (feelings of pleasure and excitement and so on), then it seems possible that the Patterned View will recommend a rule that

Rule (F) will be credited with all the good caused by such affairs. Moreover, since it only takes into account good and bad arising from those affairs that stay secret, it will not be credited with significant amounts of agent-neutral bad. But this looks wrong. Having an extra marital affair is not morally right just because, in the actual world, people often get away with it (something, we might note, recognized by traditional rule consequentialism). Call this the Getting Away With It Problem.Footnote ¹⁰

Even on its own, the Getting Away With It Problem looks like a significant challenge for the Patterned View. However, we might think that it also reflects a deeper, counterintuitive feature of that position: namely, its unswerving focus on just that possible world that happens to be actual. After all, often what happens, happens, despite what, all other things being equal, we might think should have occurred. To put it another way: it often feels like there's no justice in the world – or at least, a lot less justice than we might hope. As such, the world can look like an odd place from which to draw standards of justice. From here, then, traditional rule consequentialists might reasonably claim that the possible world the Patterned View chooses as the grounds for its moral demands looks arbitrary when compared to the hypothetical worlds they consider.Footnote ¹¹

Overall, therefore, it seems difficult to conclude that, from the mere fact that the Patterned View avoids the Ideal Worlds objection (hypothetical worlds), it has an advantage over more traditional forms of rule consequentialism. And that is leaving to one side the fact that, even if it were established, such a claim does nothing to contest the fact that the Patterned View is open to the actual Ideal Worlds objection.