I reviewed the account of moral dilemmas that Parfit poses with regard to universalisation tests in Climbing the Mountain and in this posting I am going to track the parallel treatment that he provided in Chapter 13 of the 2008 version of On What Matters. This chapter is given a title in the 2008 version that is new as it is now headed "what if everyone did that?", a question often used to crudely summarise the basic idea taken to be behind Kant's universalisation tests.
It is interesting to see how Parfit summarises the understanding he has of the question given in the chapter title. It is to the effect that what has to be asked, in testing maxims, is whether we could rationally will it to be true that everyone acts on some given maxim. However it is not merely that this is the translation Parfit gives of the question that appears to be asked when the universal law formula is related to maxims. It is also a question that Parfit considers has to include reference to alternatives. So, it is not just whether everyone could rationally will in a certain way, but a relation of this question to options that the situation includes that would be different. The way this is understood by Parfit is that the acting of everyone in accord with a "bad" maxim could be better than an alternative if the alternative were that everyone apart from ourselves were to act in a different way. This would ensure however, that the way the maxim testing procedure is going to go is through axiological measurement. An alternative to this which Parfit considers and approves of as preferable is that we ask instead whether rational willing of a maxim is better if everyone acts on the maxim in question rather than no one acting on it. But even this is complicated by the question of what other maxims people would act upon rather the one being put to the test.
Parfit argues that the Law of Nature Formula works best when it is applied to maxims of which 3 things are true, namely that it would be possible for many to act on the maxim, that the effects would be similar regardless of the number who so acted and that these effects would be equally distributed. The reason Parfit takes trouble to specify things in this way becomes clear when the topic of moral dilemmas subsequently enters the picture as these dilemmas are understood in a particular way. The dilemmas are taken from cases of game theory in which there exist what are termed "each-we" dilemmas of practical rationality in which "if each rather than none of us does what would be in a certain way better, we would be doing what would be, in this same way, worse".
Now the way these dilemmas are construed is not neutral between different ways the universalisation test could be understood. In these dilemmas there is an implicit assumption which Parfit makes explicit to the effect that what is really at issue are questions of "benefit" or "well-being". This applies to the classic prisoner's dilemmas cases which Parfit understands as "self-benefiting" dilemmas in which the whole question turns on whether self-interest is best served by doing what appears immediately to maximise it or rather by doing what would, should cooperation follow, be best overall. Similar considerations are at issue in "contributors dilemmas" that frame discussions of public goods. Parfit complicates matters in one way by introducing a variant that is not usually mentioned in these contexts. The variant concerns special attachments which are favoured by ordinary common sense morality including those formed between parents and children. These special attachments frame connections that Parfit terms M-relations and there can be "dilemmas" concerning whether we favour people in these relations as opposed to having impartial aims.
The public policy implications of these dilemmas are typically addressed as Parfit indicates by means of coercive measures intended to disincentivise "free-riding" behaviours. However in the moral arena act consequentialists favour direct responses to these dilemmas, albeit ones that have made many uncomfortable. Kantian solutions appear, by contrast, to favour universal responses on the grounds that the converse are not, as Parfit understands it, rationally willable. Kant's tests apply also not simply to "actual" cases (as do solutions to policy dilemmas) but also to possible ones. It is in so doing that they interest Parfit who thinks that Kantian ways of formulating views concerning putative possible dilemmas are capable of challenging the standard intuitions of common sense morality. An example of this concerns the connection of Kant's universalisation test to acts that appear trivial in impact when related to each person's acts but which have a large accumulative effect. So, on grounds of appeal to the Formula of the Law of Nature Parfit takes it there is a duty not to act so that the earth's resources become seriously depleted as this would not be rationally willable.
If Parfit thinks that appeal to the universalisation tests has a good sense with regard to cases where particular acts seem not to be wrong in themselves but only in their accumulative effect he nonetheless has reservations about the appeal to the Formula of the Law of Nature in other areas. In many cases, on Parfit's view, the problem is not picked out by the universalisation test since the wrongness of an act may often depend, on Parfit's conception, on how many people act in a given way. So when a certain number, call it x, acts on a given maxim, the effect of their so acting is neutral but when the number rises above this, it becomes bad. (Note again the axiological measurement procedure appealed to here.) The example given refers to a behaviour manifested by Kant himself and some of the rest of us. This would be not to have children so as to devote oneself to philosophy. If a certain number act on this maxim the result can be beneficial but if the number rises too high the result would be negative. Notice that the suggestion here of how to assess the maxim assumes, however, that the existence of people is something that can be calculated in terms of its goodness or badness, something that is not at all obvious to Kantians.
The objection that is put here is one that Parfit terms a "Threshold Objection" and Thomas Pogge is mentioned now as someone who thinks that a response to this objection would require us to move away from the Law of Nature Formula to considerations based on moral belief. This would mean not assessing the maxim given in terms of acts but instead in terms of what is believed concerning acts. Parfit is far from convinced that this response is a good one since Pogge's response appears to involve us endorsing beliefs that would, in regard to the acts, specify something untrue of them. Pogge also appeals to conditionality considerations which suggest that acts are endorsed only given certain conditions. This does involve Pogge giving concession however to the implicit consequentialism of the discussion.
In considering this claim about conditionality Parfit looks at a different kind of example such as what is at work in someone acting on the "maxim" that they should become a dentist. Clearly not everyone could become one but there is conditionality built into this maxim since the aim of becoming a dentist refers, at least implicitly to the understanding that this is a stable way of making a living. But, if this kind of case is handled well by reference to moral beliefs of the sort appealed to by Pogge, there are exceptions to it. Not only is this so but, on Parfit's demanding account, we have to consider not just maxims that anyone would endorse but any that someone conceivably "could" endorse. This involves conjuring up a figure familiar from the writings of R.M. Hare, the figure of the moral "fanatic". Such a figure adopts maxims unconditionally and hence is not dealt with by the type of move Pogge has made.
When we reach this level of account however, it appears that the type of objection that is really at work in "threshold cases" is not really a new objection after all as the figure of the fanatic makes clear that what is being considered are cases of "mixed maxims", maxims that it would be sometimes but not always wrong to act upon. And when it becomes clear that the cases in question are really of this type Parfit alters the account he gives of the way that Kantians could respond. Now Parfit suggests that "maxims" should not be considered as referring to general "policies" but instead to morally relevant descriptions of what people are doing. This leads to a revised version of the Law of Nature Formula being given: "We act wrongly unless we are doing something that we could rationally will everyone to do, in similar circumstances, if they can".
The reference to similar circumstances is supposed to cover even the case of the moral fanatic and their unconditional endorsement of maxims. Having introduced the variant of the Law of Nature Formula intended to address the renewed "mixed maxims" objection Parfit next considers a different type of objection. On this objection it would be the case that if enough people act in a certain way, these acts would have good effects but when fewer people act in this way the effects would or might be very bad. Here we have a kind of "reverse threshold" since it is the case that it is the number dropping below x that produces the problem. A case said to indicate the problem is real is that of the maxim, "never use violence" which, according to some, is one that Kant's tests should lead us to endorse but which clear cases (like those prevailing in the 1930's) count against. Game theoretic considerations of similar sort are also again wheeled on here by Parfit. The point of this reverse threshold objection is to motivate the general complaint, attributed to Christine Korsgaard, that Kant provides only standards of conduct that apply to ideal states of affairs and not to actual ones in which they might have disastrous results.
Parfit considers conditional statements of maxims again in order to show that it need not follow that maxims are endorsed that fail to take any account of what others are doing. This enables, for example, a much clearer statement of response to violent acts which allows response to the violence of others should that be manifested. Given that conditionality of maxim formation allows a response to the "ideal world objection" it is not so clear that this can be sustained. However, rather than this being the case, it could instead be true, as Parfit considers, that unrestricted retaliation to violence could be rationally willed. Since this would only be in response to actual violence undergone Parfit assumes that this type of maxim could be, mistakenly, endorsed by Kant's test though it is not at all clear to me why Parfit thinks this. If the onus passes from being unreasonably restrictive to being unreasonably permissive there still appears to be a way of thinking about thresholds that creates a problem (at least according to Parfit). These problems do often afflict rule consequentialism, as Parfit indicates. On the basis of consideration of this reverse threshold objection Parfit revises further the Law of Nature Formula so that it becomes: "It is wrong for us to act on some maxim unless we could rationally will it to be true that this maxim be acted on by everyone, and by any other number of people, rather than by no one".
This revised maxim builds in consideration of the numbers of others that act according to the maxim given. The advantage of this, on Parfit's view, is that many more maxims are now condemned so that Kant's test does not become unreasonably permissive. This revision is one that Parfit assumes further enables it to be the case that moral beliefs that can be endorsed by most are thereby capable of being pursued.