Tuesday, 29 May 2012

O'Neill and Kant on Rational Willing

In a piece originally published in 1985 and reprinted in Paul Guyer's edited collection on the Groundwork Onora O'Neill discusses the understanding of consistent willing that Kant offers with his idea of willing universal laws. It's worth going through the general argument of this piece again, not least since it stands as a corrective against views that, due to the influence of Derek Parfit, are once again current concerning the categorical imperative.

O'Neill opens the piece by pointing out that there is a tendency to discuss universality tests in terms of what everybody or somebody wants done either by or to everyone. Due to understanding what wants done in terms of what happens to be contingently wanted this way of presenting universality tests ensures that they are seen in heteronomous fashion. Kant's own view of universality testing makes no reference either to want everyone wants done or to what somebody wants done either by or to everybody. This is why the Formula of Universal Law (FUL) reads simply: "Act only on that maxim which you can at the same time will that it should become a universal law" (Ak. 4: 421). So, as O'Neill puts this, we are invited here to consider that we can will only that which it is possible or consistent to will as a universal law and no reference is made to what we would will or would find acceptable or would want. So Kant's formulation does not match those that are often given. This is why Kant is taken by O'Neill to have what she terms "an uncompromisingly rationalist foundation for ethics". I'd have to parse this notion of "rationalism" a little myself but it is, I think, clear that what O'Neill means here is that the formulation is one that moves away from desire/inclination models entirely and that this is what is distinctive about it.

Now, one of the reasons why universal law tests tend to import heteronomous elements in their discussion is to avoid charges of triviality or emptiness (as the shadow of Hegel falls over most discussion of universal law tests). Substantive conclusions are generally thought not to be available from formal tests alone. O'Neill's piece sets out to show why this view is problematic and thus to offer a basis for viewing formal universality testing of a Kantian sort as capable of responding to the challenge of providing substantive conclusions. 

The first step in O'Neill's reconstruction is to bring out that there are two aspects to how the Kantian universality test is to go and in describing these she brings out how FUL proposes to us two points. Firstly, FUL indicates reference to action on a maxim. Secondly, it restricts the types of maxim we can appeal to by reference to the capability of a maxim being one that can become a universal law. So the two elements of working out how to understand what "action on a maxim" means and how the restriction of maxims to conditions of universality is to work have to distinguished in order for the test to be understood in its complexity.

So O'Neill's analysis opens with an account of what "acting on a maxim" involves and follows Kant's basic statements of "maxims" as being "subjective principles" of action (Ak. 4: 421n). Such a point is not meant by Kant to suggest that they are "subjective" in the sense of being purely arbitrary or having any content whatsoever. Maxims are not only constrained in their form in ways we'll look at in a moment but they are also available generally in the sense that any number of agents could adopt them, they could be adopted at different times and places and refer to apparently very different actions. However, it is also the case that whilst Kant discusses action through his conception of "maxims" that it is no part of the Kantian conception to think of "maxims" as being consciously referred to always. But in a given situation there are involved in the description of the "maxim" of action references to the agent, the act and the situation itself. 

Maxims are not (pace Parfit) the same thing as "intentions", not least because "intentions" include many details that are not relevant to the consideration of maxims. Maxims are, rather, the underlying principles that guide specific "intentions" as O'Neill puts this. I'll formulate this point slightly differently to how O'Neill does but, I think, in conformity with the general thrust of her view. On my account, a maxim can be understood in general terms as formulating a kind of aim we have, one that exists whether it is consciously at work for us at various times or not and which forms the context in which we develop specific policies that are expressed in given specific actions. This understanding of maxims as stating aims is important although it does mean that it is not always transparent, simply from observing someone's actions, what maxim they are following. Nor might an agent always be able to answer if we pressed them as to their maxim since it is possible the general aim of their action was not at that given point before them even though the understanding of it as a background condition is what really made sense of their action. So we infer them as a result of understanding not specific actions in isolation but by connecting specific actions together in view of the policies involved in them in order to arrive at a sense of the basic aim in view.

Now part of O'Neill's point in this article is to use this argument to distance Kant from a general understanding of universality tests that frames the way they are put by reference to heteronomous considerations. The point here is that universality tests tend to be seen through the prism, as O'Neill puts it, of "categories of right". Rightness and wrongness are then seen primarily in terms of appraisal of outward actions. On O'Neill's account, however, this is not the prime point of using the categorical imperative in a universality test with regard to maxims. Rather than this being the point the basic question the categorical imperative is being used to put is whether the aim with which we propose to act is consistently universalizable. If this is the point of the test then it follows that what we want to know when we apply it is whether acting on this maxim will produce an act that is morally worthy and this is the way we can tell whether or not the act is one that conforms to duty. So duty is comprehended through the question of moral worth and not the other way around on O'Neill's view. 

This key point distances the understanding of the point of the categorical imperative's universality test away from understanding of outward actions towards moral worth with moral worth now emerging not as something that is merely bound by duty but rather as the central case of duty. Understanding of outward appraisal of actions would, by contrast, be derivative of this. Putting matters this way shows that the understanding of what "action on a maxim" involves is not a trivial part of the process of constructing the universality test.

A result of seeing the Kantian universality test in this way is that the general objections to universality testing that are posed by many can be seen to be faulty. So, for example, the basic problem of "maxim-fiddling" that is formulated to show that any act can pass the universality test arises because it is taken to be the point of the test to appraise outward actions and to see whether if the actions in question were performed by others a contradiction would arise. By contrast, on O'Neill's construction, the outward actions are not the primary point of the test. Another way that counter-examples tend to get presented is through focus on specific policies as the target of maxim testing rather than underlying aims, a basic failure to get the point of the reference to maxims.

The first point of the process is thus to identify maxims as aims of actions and to evaluate policies by reference to aims. It is only a secondary part of the process to appraise outward actions. A major plus of this reading is that it brings the first two parts of the Groundwork closer together than they are often seen as. The account of moral worth in the first part of the Groundwork is frequently seen as if it had little to do with the account of the categorical imperative in the second part whereas O'Neill's view leads us to seeing the categorical imperative instead through the prism of moral worth. It is thus the consequence of her argument that we should basically apply the universality test to our own aims of action as it is we who are adopting maxims. 

The second part of the process opens now we have an account of what "acting on a maxim" involves. This second part requires testing the maxim in terms of its consistency with universality. O'Neill famously formulates two types of inconsistency: one within an agent's maxim (contradiction in conception) and one between either the different policies that follow from the aim or between the policies and the aim (contradiction in the will). The first type is here termed by O'Neill conceptual inconsistency and the second type volitional inconsistency. There are different ways that maxims can express either of these types of inconsistency.

The most basic way a maxim can be inconsistent is by expressing an aspiration which is, for some reason, impossible. This is so if, for instance, it involves mutually incompatible aspirations. This would prevent action on the underlying maxim from being possible and would point to the emergence of action that is seriously disjointed as a consequence. However, the point of my distinction between policies and aims comes out clearly when we consider how O'Neill formulates this point: "A non-universalisable maxim embodies a conceptual contradiction only if it aims at achieving mutually incompatible objectives and so cannot under any circumstances be acted on with success". It is clear that the underlying aim is the problem here and this is what a maxim is basically concerned with. The policies come into conflict with themselves as they are part of an aim that is intrinsically impossible to realise.

Maxims are principles of action in that they give us principles that we aim to realise and this aim is what is meant by saying that we have willed the maxim. The conceptual contradiction emerges when the willing in question incorporates something in its aim that is not possible. The more interesting case, in many respects, however, concerns volitional inconsistency. It is with regard to it that O'Neill is led to an account of rational action that is particularly rich. Kant describes hypothetical imperatives as involving a kind of analytic connection between means and end assuming that reason has decisive influence on action. O'Neill terms this the Principle of Hypothetical Imperatives (PHI) and says that it means that agents "intend any indispensable requirements" for the achievement of their aims. However, whilst Kant's text in the second part of the Groundwork appears to rest with this point, O'Neill articulates a broader view of rational aiming (called by her rational intending but I think this is misleading). The broader account leads to her setting alongside the PHI a list that is unlikely to be complete but which extends the general discussion of rational volition further.

Included amongst O'Neill's family of principles of rational aiming are the following: a) a requirement of rationality to direct policies that are not merely concerned with indispensable or necessary means (as with the PHI) but also with sufficient means for achieving the aim; b) adding to sufficient means the seeking of such means when they are not available; c) including in one's policies all necessary and sufficient components of what is aimed at; d) ensuring that specific policies are mutually consistent; e) ensuring that the foreseeable results of specific policies do not conflict with the underlying aim.

O'Neill understands the point of the family of principles as one of bringing together what she calls "surface and underlying intentions" but which I would term bringing together policies and aims. Policies must express commitment to acts that provide either the means to or components of aims and must not undercut the aims. The fact that the policy not undercut the aim is central here and points to a basic standard of action. 

Having clarified this point O'Neill returns now to the universality test and states that her view is that the basic intuition expressed in it is a non-egoistic one of not singling ourselves out for special consideration or treatment. This is clear when Kant talks about how we often "make exceptions" of ourselves when we attempt to formulate maxims and O'Neill understands this point to mean that FUL includes in its basic reference a notion of a plurality of interacting agents. The consideration of maxims is one in which we are understanding ourselves to adopt a maxim that meets the condition of being able to be adopted by others and maxims, as we will see in a moment, often are concerned with relations to others. 

Now when the universalisation of maxims is presented in the Groundwork Kant moves from FUL to the Law of Nature formula and after referring to the latter he states the two types of inconsistency (Ak. 4: 424). Contradictions in conception often would not involve a contradiction if we did not attempt to universalise them so it is the process of putting them to the test of universality that reveals the problem with them. A clear case of this occurring is with maxims of deception which involve no incoherence so long as they are not universalised but which do when they are. With regard to the examples of contradiction in conception the distinction between aims and policies is particularly important. 

Maxims that lead to volitional inconsistency state no manifest incoherence even when universalised. So what they conflict with is one of the family of principles of rational volition that O'Neill spoke of earlier. The most obvious case is where they conflict with PHI which Kant takes to apply to the case of non-beneficence. The maxim of neglect of talents, by contrast, is one that O'Neill thinks contravenes not merely PHI but also the second requirement of adoption of sufficient means for realisation of aims. And she has similar arguments that invoke her criteria of rational action with regard to other types of maxims. The point of the cases of contradiction of the will is that they show that a number of maxims require, for their realisation, stable relations with others as a necessary condition. The general argument thus shows that a Kantian argument for a universality test that is grounded on considerations of formal consistency and rationality can ground substantive requirements without needing to refer to notions of desire and preference.


Yeah, OK said...

Thanks for this. It's inspired a post of mine (in which I admittedly may be missing something)


Gary Banham said...

Thanks for your reply and I hope you do't mind if I reply here rather than on your own blog. Your posting suggests that the maxim of deception is specific in focus, not general but there is precisely where my distinction between *aim* and *policy* allows me room to reply. The specific *policy* is, as you say, related to the particulars of the case which lead me to aim to deceive the particular person but this *policy* is formulated as a means to a general end and it is within this context that the policy is taken to fulfil the *aim* in question. The *aim* is what the maxim expresses and the Kantian test is one in which the worthiness of maxims is tested by universalisation. When the aim is universalized it becomes one of assuming that reaching ends by means of deception is permissible and this leads us to the desired inconsistency.

Yeah, OK said...

Hi Gary,

Thanks for this. I'm having a hard time responding because I'm not entirely sure what, in the deception case, is the content of the agent's "aim" and "policy". Could you fill in those blanks?

Gary Banham said...

OK: the way I see it, the *aim* of action is to achieve a certain end by a certain means taken to be permissible in relation to the end. By contrast, the *policy* is specific in its application in terms of range and content (so it refers to the given person in the given situation). In applying the universality test you aren't asking about the given situation in its specificity but rather about the generalised understanding of how means and ends permissibly relate (which is the *aim*). Does this help?