Allison and Kant on Rational Agency

In my last posting on Allison's commentary on the Groundwork I looked at Chapter 5 which was a loose collection of material roughly relating to the concluding sections of Groundwork I. In this posting I am going to look at Chapter 6 which, conversely, is more tightly focused but is somewhat preparatory to a serious focus on the argument of Groundwork II.


In this chapter Allison is looking at Kant's account of rational agency and how it relates to his description of imperatives. One of the problems that Allison has posed insistently since the opening of his commentary concerns why Kant's argument did not move simply from the discussion of common moral cognition in Groundwork I to the "deduction" of the categorical imperative in Groundwork III? The basic reason that this chapter opens with is that Kant needs to establish the absolute necessity of the supreme principle of morality and that this is the central task of Groundwork II. However there is also a second question about Groundwork II that concerns its internal structure and the apparent multiplicity of formulas of law that it gives. Like Paul Guyer, with whom he otherwise differs, Allison takes the view that the formulas are parts of a progressive description that Kant is giving in Groundwork II, the description, according to Allison, of the concept of finite rational agency.


The account of rational agency opens, according to Allison, at Ak. 4: 412 where Kant speaks of how only rational beings have the capacity to act according to the representation of laws as opposed to simply acting on these laws themselves. What kind of law is this? Allison considers the argument that it is a moral law and adduces two factors in support of such a view: firstly, Kant describes only the categorical imperative (among practical principles) as a law and, secondly, since maxims are freely adopted by agents this has the virtue of placing freedom at the heart of Kant's account. However Allison decides against this reading and prefers the view that what is meant by "law" here is, rather, objective practical principles (both moral and instrumental). 


The key point about the reference to laws in rational agency is, however, that it introduces the notion of the "will". Objective principles as necessitating for wills are commands of reason and the formula of their commands is an imperative (Ak. 4: 413). Once we have this notion of an imperative we can divide the genus of this idea into its sub-species of hypothetical and categorical. To command categorically is to command independently of any given end. 


Allison reviews the contrast Kant draws between two forms of hypothetical imperative in the Groundwork, distinguishing, as he does there, between imperatives of skill and pragmatic imperatives. Further Allison states that hypothetical imperatives are all grounded in an analytic principle with regard to willing, a principle stated by Kant at Ak. 4: 417: "Whoever wills the end, also wills (insofar as reason has decisive influence on his action) the means that are indispensably necessary to it that are in his power". This statement is described by Allison as the "grounding principle" (GP) of hypothetical imperatives but it is notable that it does not contain any sense of inclusion of an "ought". 


Two conditions are introduced on the apparently analytic statement GP by Kant which are that the agent must have knowledge that the end in question could not be attained apart from a given course of action and that the agent must fully or completely will the end in question. As Allison points out, the second of these conditions is important given that it is all too common for agents not to fully will to attain a given end. On the basis of these further conditions Allison fills out the "grounding principle" of hypothetical imperatives further than Kant explicitly did and arrives at what he terms GP1. This has the form: "If A fully wills E and knows that M is indispensably necessary for E and M is in its power, then A will M". 


However there is still a problem with how we move from GP1 to the statement of the hypothetical imperative given that the latter contains an ought-operator and the former does not. Normativity will require some form of prescription which has yet to be stated. As Allison indicates, however, this element of prescription is often not explicitly stated in a law as when a simple traffic law states what one must do, it does not also say that you ought to do what you must do. Ought-propositions are addressed to finite rational agents and such agents also require an understanding of how ends relate to means. If an agent simply thinks an end in relation to means they have a hypothetical connection between the two that stands under the assumption of desiring the end in question. It is the importation of this that brings in the ought-operator.


What Kant takes to be analytic is thus not hypothetical imperatives themselves according to Allison but rather the practical propositions that correspond to them. Since the volition of an agent is presupposed in the statement of a hypothetical imperative and GP1 describes the means of attaining them then this is sufficient to account for the possibility of such imperatives. However, in addition to the generic description of such imperatives Thomas Hill has further argued that there is a general form for all hypothetical imperatives that can be expressed as the Hypothetical Imperative, of which specific imperatives are given examples. It could be taken to abstract from given specific ends in favour of the general way in which instrumental end-setting can be given. Allison rejects this idea though he does so very swiftly without examining the details of Hill's argument and the swiftness of this makes his rejection unconvincing.


Key to the notion of an imperative is constraint so that accounting for an imperative is to describe the grounds of constraint it imposes. With hypothetical imperatives the constraint is comprehended in relation to the end set but categorical imperatives are ones that command, in their unconditionality, move beyond this type of constraint. The necessitation of such categorical imperatives is not grounded in any given volition. However the generic possibility of categorical imperatives is not the province of the argument of Groundwork II but, instead, of Groundwork III.


Given this point Allison moves instead to looking at the derivation of the categorical imperative that is carried out in Groundwork II which culminates in the statement of its formula at Ak. 4: 420-21 and which is glossed in terms of an ought-operator that refers to consistency of maxims in relation to universal law. In the Groundwork I argument the universal law formula was grounded on the relation of imperatives to the good will and the conformity of imperatives to the good will was a sufficient reason for rejecting anything that failed to conform to universal law. The content of the law was filled by reference to the pre-given maxim's of the agent.


In Groundwork II, by contrast, the derivation of the categorical imperative is argued by Allison to follow from analysis of the concept of an imperative with the law being the source of the constraint on volition that must be operative in an imperative. The necessity that is built into the imperative is also assumed to be grounded on the reference to law in general and determinate content to be removed from the law. This leaves us with the simple conception of conformity to law as such. However, the general worry with this argument can be said to have the form that, whilst apparently stating something purely formal, it arrives at a principle with content and that the means by which this is done is unclear. Allison's response refers to the unconditionality of the statement of the law as that which provides us with a general possibility of content. Or, it does so "just in case" the agent takes their maxims to be willed also as universal laws. Key to the whole chapter has been the simple statement that rational agents act according to their representation of laws and the demand that an unconditional imperative provide prescriptive force independently of reference to ends. 

Teleology and Practical Reason

The recent postings I have done have been focused on bringing into view a sense of practical reason. One of the consequences of the relatively basic conception advanced to date is that the notion of practical reason, if viewed in terms of the general structure of hypothetical imperatives, is one in which the notion of "ends" plays the key guiding role.


I want to first indicate some provisional points about the structure of the connection between hypothetical imperatives and teleology prior to raising a larger framework for further investigation of the reasons for thinking about practical reason in a teleological way. 


Hypothetical imperatives generally fit an instrumental model of rationality in which means and ends are adjusted to each other in order to attain a desired objective. Kant describes hypothetical imperatives in two ways though the nature of them is worthy of much more extended consideration on a future occasion. The two general ways are in terms of "skill" and "prudence".


The conception of hypothetical imperatives in terms of "skill" connects them to the provision of more or less sophisticated technical ways of achieving an end. Any kind of action or activity that requires the varied use of materials in order to produce a result will involve degrees of skill in both the relationship between the materials and the judgments concerning the ways of achieving the outcomes. This broad conception of "skill" indicates that everything here that requires instrumental relations tends to incorporate a form of "skill". This will apply to everything from development of trades or talents to the running of governments or the waging of wars. Hence it takes up a large portion of activity.


Contrast this now with "prudence". In developing talents we need to have an eye for what will be in demand so that we don't waste our time. This is prudential as also is avoiding certain kinds of people or activities, either of which might impair our chances of carrying out other activities already decided upon. So prudential reasoning is the province of a kind of judgment, one that tends to get sharpened by experience. Like the reasoning at work in "skill" it is a form of instrumental relation. I'll return generally on another occasion to questions about these relations but for now it will suffice to indicate an at least provisional relationship between hypothetical imperatives and instrumental relations.


It is clear in addition that the development of activities in accordance with hypothetical imperatives requires a sense of purposiveness. Hence hypothetical imperatives are generically capable of being seen as part of a teleological pattern of life.


Having made this relation between hypothetical imperatives and teleology relatively perspicuous the next step is to relate teleology in a more general sense to practical reason whilst bracketing for now further consideration of hypothetical imperatives. There are two kinds of way in which a general relationship between teleology and practical reason tends to get understood, ways which are on distinct levels from each other. One is through a theory of moral psychology that views practical reasoning as primarily teleological. The other is a meta-ethical stance concerning the types of considerations that are relevant to assessing right acts. The first is what is generically termed the "Humean" notion of moral motivation, the second is the consequentialist view of moral appraisal.


Taking these in turn, the Humean notion of moral motivation is an analysis of the psychological preconditions of being moved to act in certain ways. The basic claim of the Humean theory is generally understood to consist in a non-cognitivist view that an ethical belief is in itself insufficient to motivate action and that an affective state has to also be activated for someone to act in accord with a belief. Due to this general claim the teleological notion at work in motivation for the Humean is one that is taken to cut against a pure notion of practical reason.


By contrast to the Humean picture that thinks of teleological connections as the means of explanation of how we are motivated to act, the consequentialist view concerns the means by which action is justified (not necessarily how it is motivated though it may also claim this). Consequences are the means of assessment of actions and the conception of teleology is thus an instrumental one of how ends are best reached and how the "right" ends can be maximised.


The connection between means and ends in both of these pictures seems to have the pattern of an instrumental relation and in this respect to be of a piece with what we have uncovered concerning hypothetical imperatives. Due to this there is a general view that both the Humean conception of motivation and the consequential notion of ethical justification are inherently non or even anti-Kantian views. Certainly the consequence of the Humean view of motivation is that there can be no pure practical reason and in this important respect it is at variance with the Kantian view. It remains true, however, that there does need to be a Kantian picture of how we can be led to act in accordance with the moral law. Such a picture thus has to provide some kind of response to the Humean notion of motivation.


The consequentialist picture of justification, by contrast, whilst widely taken to be at variance with the Kantian picture of justification, is not universally so regarded. Rawls and his followers take it to be paradigmatically the view of morality that Kantians must reject and to which Kantian theory is a response. However, there has been developed a notion of "Kantian consequentialism" that aims to show there is not only no necessary conflict between Kant's moral theory and consequentialism but that they can even be combined.


The setting out of responses to both the Humean and the consequentialist views is essential to clarifying what the Kantian argument concerning the relationship between teleology and practical reason should be and will be part of what is at issue in many future postings.

Intelligence and Hypothetical Imperatives

Following on from the last posting I was wondering next how to relate Korsgaard's view of intelligence to the conception of hypothetical imperatives as a step to answering the question I raised in conclusion concerning maxims. Recall, on Korsgaard's conception, that a "maxim" (or maxim-like structure) can apply in cases of intelligent but non-rational behaviour in response to the environment.


This case happens when there is learning concerning action that requires normative appraisal of the right thing to do. So, in the case Korsgaard gives, the act of moving in the direction of food (as opposed to the automatic responses of the body such as salivation) constitutes an acceptance by the organism in question of a principled response to the stimuli in question. One of the reasons for taking this to be the case is that the smell of food can, given the right pattern of stimulus and response, be one of avoidance. So, if the smell of honey is followed by the stinging of bees, the organism could well learn the inappropriateness of attempting to consume the honey. These cases of intelligent behaviour don't require self-conscious reflection on the principles as principles but they do require acceptance of the appropriateness of the action (which is taken as a principle by Korsgaard).


The point I concluded with in the last posting was that if this is type of action is effectively of a maxim-like structure and if such behaviour can also characterise beings who do possess reason then what is the means by which rational beings can be moved from acting in accordance with intelligent "maxims" to rational ones?


The rational action requires not merely acceptance of the appropriateness of the action but self-conscious affirmation of the principle that underlies it which is now explicitly related to as a principle. A key consequence of the shift is that we move from "instincts" to "reasons" where the latter are here taken to constitute grounds for action in the sense of providing action with an end. There are, as will be discussed on future occasions, numerous questions that can be raised about end-oriented behaviour. 


For now, I'll simply adopt Korsgaard's very compressed conception of ends in which we do act-A for the sake of end-E. The most basic conception of this is that the notion of the end is something that we desire to attain and we adopt means that will enable this to be the case. Adopting such a view will be sufficient to give us the notion of hypothetical imperatives.


A hypothetical imperative is described in the second part of Kant's Groundwork as the capacity to act in accordance with the representation of a law or, in other words, acting on the basis of willing. Hypothetical imperatives are so called because in such a case we act on the view that a certain means will be the basis for attaining a certain end and this connection requires viewing something in a hypothetical connection.


Now the difference between what is at issue in a hypothetical imperative and what is occurring when we observe merely intelligent behaviour is that the relationship between means and ends in the former case is one we have explicitly postulated and taken as desirable. The "incorporation" of which Allison speaks and on which Korsgaard's model trades has been understood here to be one we are open and articulate about.


If this relationship between intelligence and hypothetical imperatives is correct then it follows that there is a continuity between them. Hypothetical imperatives are something like the rational expression of basic intelligent behaviour. Whilst this means there is a difference in kind between hypothetical imperatives and intelligent behaviour in the sense that in the former cases there is self-conscious affirmation of principles as principles, the types of principle involved in hypothetical imperatives are still of a piece with the types of principles given to intelligent beings.


So to view intelligent behaviour as having a maxim-like structure is to see it as having a form of much the same kind as that of hypothetical imperatives. In a sense, then, the question I raised at the end of the last posting can now be re-phrased. The move from intelligent maxim-like structures to explicit maxims is one in which the nature of what is affirmed in the maxim does not have to be explicitly changed, only the relation to the principle (in terms of articulacy) does. Hence if there is here a continuity between intelligent and rational behaviour inasmuch as they are seen as goal-driven then the real question rather concerns the basis for the view that there is such a thing as pure practical reason. Only on the conception that there is a pure practical reason does it emerge that there is something about rational beings that is intrinsically and deeply specific that sharply distinguishes them from intelligent beings. This helps to explain the Humean view of "human nature".