1. The object of moral evaluation

In the Groundwork, Kant uses the term ‘maxim’ several times before he provides the following note: ‘A maxim is the subjective principle of willing; the objective principle … is the practical law’ (G, 4: 400 n*, p.16). Beyond connecting maxims to acts of willing and contrasting their subjectivity to the objectivity of practical laws, he does not explain how we are to understand these principles. Perhaps Kant thought that he had already adequately clarified this matter with his opening argument and examples. As Henry Allison explains, Kant’s goal in the first section of the Groundwork is to provide ‘a clarification or making explicit of what is supposedly implicit in a shared, pre-philosophical, understanding of morality’.Footnote ⁴ Kant presents a series of examples, assumes that his readers will share his common-sense moral judgements, and then draws conclusions about the idea of a good will. We will use the first two examples to identify what must be included in a maxim if it is to serve as the primary object of moral evaluation.

The first and best-known of these examples is that of the shopkeeper (or merchant). Kant is interested in whether we can conclude that a shopkeeper who does not take advantage of his customers has a good will (i.e., is morally virtuous). He writes:

Kant identifies three different grounds on which the merchant might engage in the outward behaviour of charging everyone (even an inexperienced child) a fair price. First, the merchant might charge a fair price based on ‘an immediate inclination towards his customers, so as from love, as it were, to give no one preference over another in the matter of price’ (G, 4: 397, p.13). Kant dismisses this possibility as implausible in this case. Second, he might refrain from taking advantage in order to cultivate a reputation for honesty which will be better for his business in the long term. The third possibility is that he charges a fair price because he is committed to ‘a principle of honesty’ and recognises that doing anything else would be wrong (unjust). Having identified the merchant as ‘prudent’,Footnote ⁵ Kant asserts that in this case, ‘the action was done neither from duty, nor from immediate inclination, but merely for a self-interested purpose’ (G, 4: 397, p.13).

Some readers have taken the shopkeeper example to suggest that Kant believes we only act with a good will when we act contrary to our inclinations or self-interest,Footnote ⁶ but Kant would later repudiate this interpretation.Footnote ⁷ His main point is simply that common-sense morality registers a relevant difference between the merchant who charges a fair price because he values that action as contributing instrumentally to his long-term self-interest and the merchant who charges a fair price because he recognises that morality requires it regardless of any likely further benefits. In neither case is the outward behaviour for which they are responsible – what Kant calls their deed (G, 4: 420 n*, p.33; MM, 6: 223, pp.18-19) – unjust (or contrary to right). That is the point of saying that the customers are served honestly. However, a judgement of virtue cannot be based on the external deed alone. Common-sense morality requires that we also consider the end that the agent aims to achieve.

Kant holds that every action has an end (MM, 6: 385, p.159). In the case of instrumental actions, the deed is performed in order to bring about some further and independent goal. As Kant describes him, the (merely) prudent merchant charges a fair price (his deed) because he believes that doing so will ultimately benefit his business (his end). Not all actions are instrumental in this way, however. Some deeds are done for their own sake, in which case they are their own ends. The merchant whom we imagine charging the fair price because he recognises doing so to be a requirement of justice does not act for the sake of some further and independent end. Charging the fair price is itself his goal, regardless of any further consequences. If maxims are to be the primary object of moral evaluation, they must specify not only the deed but also the intended end (in particular, distinguishing whether the deed is done for the sake of some independent end or for its own sake), and Kant is explicit on this point: ‘every maxim of action contains an end’ (MM, 6: 395, p.168).

The shopkeeper example might suggest that all actions that are done for their own sake are examples of a good will while all instrumental actions are not, but Kant’s second example shows that this is not correct. Kant claims that ‘to preserve one’s life is one’s duty’, but in addition, ‘everyone has an immediate inclination to do so’ (G, 4: 397, p.13). (This inclination corresponds to the dismissed possibility that the merchant charged a fair price because of a direct inclination of love that he felt towards his customers.) But, Kant says, for ‘the greatest part of humanity’ preserving one’s life is done merely ‘in conformity with duty, but not from duty’ (G, 4: 397-398, p.13). To have a clear case of a good will, we must consider an ‘unfortunate’ person for whom ‘adversities and hopeless grief have entirely taken away the taste for life’ – in other words, one who no longer has that immediate inclination – who ‘yet preserves his life, without loving it, not from inclination, or fear, but from duty’ (G, 4: 398, p.13). Once again, the point is to contrast two cases in which the outward behaviour is the same, in this case preserving one’s life, but for which, Kant claims, common-sense morality makes different moral assessments. Unlike the shopkeeper example, here the contrast is between two cases in which the act is performed for its own sake, non-instrumentally. In the first case, which Kant assumes is more typical, the act of self-preservation is taken to be worth doing for its own sake because one has a direct inclination towards it, while in the second case it is a recognition of the demands of morality which motivates the act. Preserving one’s life from inclination is not wrong, of course, just as the shopkeeper who acts prudently does not act unjustly, but Kant assumes that we share the judgement that such an action does not exhibit the moral virtue of the second case. Since common-sense moral judgements distinguish these cases, maxims must register the difference as well, so it is not enough simply to include the end served by the act. In addition, the maxim must specify what aspect of the act (or end) the agent takes to be valuable, or what Kant calls the ‘incentive’.Footnote ⁸

The precise syntax of maxims is not important for us. What does matter is that the maxim includes both the end for which the action is performed and the incentive. In the case of actions undertaken instrumentally, since the end is distinct from the action, the end can typically be specified in a way that directly reveals the incentive. If we say that the prudent merchant charges a fair price in order to maintain his reputation, the inclusion of that end in the statement of his maxim also identifies his incentive since it can be assumed that the description of the end identifies the relevant sense in which he takes it to be good. In the case of actions done for their own sake, however, it is crucial that the maxim identify why the deed is taken to be worth doing for its own sake. In particular, the maxim must specify whether it is taken to be good because of a direct empirical inclination or because of a recognition that the action is required, either by morality itself or by some other social norm. In a crucial passage from Religion within the Boundaries of Mere Reason (1793), Kant indicates that maxims include incentives: ‘freedom of the power of choice has the characteristic, entirely peculiar to it, that it cannot be determined to action through any incentive except so far as the human being has incorporated it into his maxim’ (R, 6: 23-4, p.49; cf. L-Met, 29: 900, pp.266-7; MM, 6: 218, p.23).

There is some controversy in the secondary literature concerning whether Kant thinks that there is always a maxim whenever we act.Footnote ⁹ There are some passages where Kant seems to suggest that acting on maxims is a virtue of character – an achievement that we should strive for, rather than something that necessarily characterises all of our actions. We will return to this issue below when we discuss strength of character, but the previous quote from the Religion seems to indicate that actions always have a maxim,Footnote ¹⁰ and Kant makes the point explicitly in his Lectures on Metaphysics (1792-3): ‘as a freely acting being, a human being actually cannot do anything without the will – he acts always according to maxims even if not universally’ (L-Met, 28: 678, p.380).Footnote ¹¹ Further, this is consistent with the deeper connections to Kant’s account of causality and the transcendental idealism of the Critique of Pure Reason that we will explore below.

To summarise this section, we can say that in order to serve as the primary object of moral assessment, maxims must specify not only the deed performed but also include the end of the action (which may be the deed itself or a distinct goal, in the case of instrumental actions) as well as the incentive (what aspect of the action or end the agent takes to be valuable). However, notice that in the discussion of these examples, Kant has simply stipulated the maxims on which various illustrative actions occur. This is important to note because Kant holds that we never directly observe a maxim (R, 6: 20, p.46). Indeed, we never have an impression of even our own maxims, and therefore, even ‘with the acutest self-examination … it cannot be inferred with certainty’ that we acted from a moral maxim rather than merely in conformity with morality and from some ‘covert impulse of self-love under the mere pretense of that idea’ (G, 4: 407, p.22). In the examples throughout the Groundwork, Kant suppresses these epistemic worries by treating the maxims simply as given.

2. Actions and causation

In addition to serving as the primary object of moral evaluation, maxims are essentially connected to the causality of our will. In fact, it is precisely by giving ourselves maxims that our will exercises its causal powers. For Kant, having a will just means having a certain kind of causal power:

To clarify how maxims are involved in the causality of the will, we must first examine Kant’s parallel account of natural causality.

In the introduction to the second edition (the B-edition) of the Critique of Pure Reason, Kant identifies the key to his understanding of causation:

Kant agrees with Hume that we never have an impression within experience of causation or, more generally, of any type of necessity. As he puts it in the Prolegomena: ‘The concept of cause contains a rule, according to which from one state of affairs another follows with necessity; but experience can only show us that from one state of things another state often, or, at best, commonly, follows, and it can therefore furnish neither strict universality nor necessity (and so forth)’ (P, 4: 315, p.108; cf. CPR, A243/B301, p.343). But Kant argues that the concept of causation must be presupposed in order to make any empirical judgements of an objective reality. This is because we must distinguish between the subjective temporal order in which we experience events and their objective temporal order. To summarise Kant’s pair of examples, suppose I observe first the roof of a house and then turn my gaze to the ground below it. There is a subjective sequence of my perceptions: first the roof and then the ground. However, I ‘could also have begun below and ended above. … In the series of these perceptions there was therefore no determinate order that made it necessary when I had to begin in the apprehension in order to combine the manifold empirically’ (A192-3/B237-8, p.307). In contrast, if ‘I see a ship driven downstream[, m]y perception of its position downstream follows the perception of its position upstream, and it is impossible that in the apprehension of this appearance the ship should first be perceived downstream and afterwards upstream’ (A192/B237, p.307). In contrast to the sequence of perceptions of the house, in this case I am attributing an objective ordering to this sequence.

In attributing an objective temporal order, we implicate counterfactuals and therefore implicitly make modal judgements of necessity. If I had (counterfactually) looked first at the ground and then the roof, I would first have seen the ground and then the roof. But if I had (counterfactually) looked downstream and then upstream, I would not have seen the ship in those locations in that order. In other words, in the case of an objective temporal sequence the order is necessary, or, as Kant puts it, ‘one thing (that which happens) follows that of the other (which precedes) in accordance with a rule’ (CPR, A193/B238, p.307) This rule, which must be assumed to exist in order to attribute an objective time sequence to empirical events, is none other than a law of cause and effect that states that an event ‘is determined in time necessarily and in accordance with a rule by something that precedes it, [and] consequently the relation of cause to effect, is the condition of the objective validity of our empirical judgements with regard to the series of perceptions, thus of their empirical truth, and therefore of experience’ (A202/B247, p.312).

Kant takes this argument to rescue the concept of cause from Humean scepticism. It ‘restores to the pure concepts of the understanding [such as cause] their a priori origin, and to the universal laws of nature their validity as laws of the understanding … [This shows that the pure concepts of the understanding are not] derived from experience, but that experience is derived from them, a completely reversed type of connection that never occurred to Hume’ (P, 4: 313, p.106). Even if fully successful, however, this argument shows only that there is some causal law necessitating each event, not that we know what the law is: ‘Particular laws, because they concern empirically determined appearances, cannot be completely derived from the categories, although they all stand under them. Experience must be added in order to come to know particular laws at all’ (CPR, B165, p.264). Nor does this argument show how we are to use experience to identify natural laws. In fact, experience, being finite, will be compatible with indefinitely many possible laws: ‘in natural science there are an infinity of conjectures in regard to which certainty can never be expected’ (A480/B508, p.505). Rather than establishing any particular causal law, Kant takes his argument to establish a higher-order principle, which he states in the Second Analogy as: ‘All alterations occur in accordance with the law of the connection of cause and effect’ (B232, p.304).

To return to Kant’s ship example, we can ask why it is necessary that the ship is downstream only after it is upstream: what causes, and therefore explains, the ship’s movement downstream? The relevant common-sense causal law is not difficult to find, at least since Kant describes the ship as being ‘driven downstream’. Presumably it is something like this: when the current flows (the cause), floating objects are carried in the direction of the current (the effect). Given the current, it was not an accident that the ship moved downstream, and this common-sense law shows why its movement was necessary. The judgement that the current caused the motion of the ship carries with it necessity and therefore supports a multitude of counterfactuals: if the ship had one additional crewmember on board, it still would have moved downstream; if it had been painted red, it still would have moved downstream, etc.

However, despite Kant’s talk of ‘strict’ universality and necessity, the counterfactuals that these causal claims support are obviously not unlimited. Some counterfactuals would disrupt the ‘necessary’ sequence of events even if the same cause (in this case, the current) remained in place – if the wind were sufficiently strong in the upstream direction, if the anchor had been dropped, etc. When a presumptively valid causal law does not hold in a particular case, we have a choice whether to treat this as a full refutation of the law or as indicating a previously unrecognised limitation on the law’s applicability – in effect, invoking a ceteris paribus clause. In the latter case, Kant thinks, we are led to search for some other more comprehensive law that encompasses it but is not limited in the same way.

When an object falls under multiple causal laws, Kant says it has multiple causal powers, and we are led to search for a ‘fundamental power’ that unifies them all under a more comprehensive causal law. We can posit ‘a single radical, i.e., absolutely fundamental, power. But this unity of reason is merely hypothetical. One asserts not that such a power must in fact be found, but rather that one must seek it for the benefit of reason’ (CPR, A649/B677, p.594). In other words, such a fundamental power serves as a regulative ideal. And as reason searches for a unified causal power to explain subordinate causal powers, it aims to find ‘where particular natural laws stand under more general ones’ (A650/B678, p.594). So, in addition to the ‘law’ of every event having a cause, there is also a regulative ideal of a single causal law that subordinates and unifies all of the more particular causal laws that apply only in limited circumstances.

This sequence of identifying limitations (conditions) on laws and then searching for more comprehensive laws to remove the limitations follows a familiar Kantian pattern, and we will see it repeated with maxims. He describes the general pattern as follows:

Although each causal judgement has limiting conditions on its validity, reason, in its promotion of the search for understanding, seeks to discharge these limiting conditions. This forms the basis for Kant’s diagnosis of the various ‘antinomies of pure reason’, of which the third is most directly relevant here (CPR, A444-51/B472-9, pp. 484-9). From the principle that every event has a cause, we can posit a sequence of causes, each conditional on the temporally previous event. If we then treat the entire sequence as a single temporally extended event, our principle requires that the entire sequence must have a cause (with which all conditions are discharged and the ‘unity will be completed’). This first cause cannot itself be caused or else we have not yet encompassed the entire sequence. But the existence of a first cause – what Kant calls freedom or ‘absolute causal spontaneity beginning from itself’ (A446/B474, p.484) – would contradict our principle that every event has a cause.

Kant’s resolution of the antinomy, well-known and fundamental to his transcendental idealism, is that the postulated totality is not a possible object of experience and therefore transcends the appropriate use of the concept of cause within theoretical philosophy (CPR, A642/B670, p.590). Although we can never have an experience of a spontaneous first cause, that alone cannot establish their non-existence. And by providing a critique of pure reason that limits theoretical reason to the objects of possible experience, we avoid a possible contradiction between theoretical reason’s apparent rejection of freedom of the will (as spontaneous cause) and practical reason’s dependence on this idea. As Kant puts it, this limitation on theoretical reason ‘removes an obstacle that limits or even threatens to wipe out the practical use of reason, [and therefore] this critique is also in fact of positive and very important utility, as soon as we have convinced ourselves that there is an absolutely necessary practical use of pure reason (the moral use), in which reason unavoidably extends itself beyond the boundaries of sensibility, without needing any assistance from speculative reason’ (Bxxiv-xxv, p.114).

Pulling together the elements from this section, we note that since all causation involves a necessary relationship between cause and effect, all causation involves laws which state these relationships. Since a will is a kind of causality, actions that an agent wills can also be described by laws. However, the actions of a free will, in which an agent makes herself the cause of her deed, are independent from alien causes and so cannot be described by laws of nature. They are, rather, her maxims. That such free wills exist cannot be shown by theoretical reason, but equally, their existence cannot be ruled out. Their existence is, however, simply presupposed in our exercise of practical reason; we cannot act ‘otherwise than under the idea of freedom’ (G, 4: 448, p.57; cf. MM, 397, pp.169-70, quoted below).

I suspect that part of the confusion regarding maxims and causal laws is due to the many different senses in which Kant uses the term ‘law’.Footnote ¹² Kant argues, of course, that our actions are properly governed by the categorical imperative, which he often calls ‘the moral law’. And in addition to particular laws of nature, sometimes Kant speaks of ‘The law of nature that everything that happens has a cause’ (CPR, A542/B570, p.538). Maxims are distinct from all of these other laws in that they are subjective in the sense that their scope is limited to a single agent. When I act, I make myself – not you – the cause of some end by giving myself a maxim. Of course, the categorical imperative demands that I act only on maxims that can be universalised – that others could also adopt those maxims on the basis of their practical reason, and perhaps they would act on them in the Kingdom of Ends. If a maxim were adopted universally, it would then be universal in scope and would, in effect, be a law of nature (albeit one from which individuals could depart), and this is reflected in one of Kant’s formulations of the categorical imperative: ‘so act as if the maxim of your action were to become by your will a universal law of nature’ (G, 421, p.34). But whether it could be universally adopted in this way or not, a maxim is simply a causal law, limited in scope to the agent who is responsible for the deed that the maxim explains.Footnote ¹³

3. Attributing maxims

Let us return to the shopkeeper and consider what is involved in attributing a maxim to an agent rather than merely stipulating one for expository purposes. Such an attribution will be an exercise of theoretical reason as we seek to explain some empirically observed (or observable) action. This will involve showing that the action was not accidental, but in some sense necessary, by identifying a modally robust principle for which the agent is responsible. Identifying the maxim on which the shopkeeper acted shows him to be responsible for causing the child to be charged a fair price. Suppose that we observe a single instance of the merchant charging a child a fair price. What we directly observe is his deed, not his end or his incentive, and so not his maxim. By attributing the deed to the merchant, we are saying that it was not an accident that the child was charged a fair price, but rather that the merchant was responsible for causing that to happen. Causation implies necessity, so within limits, had things been different, the shopkeeper still would have seen to it that the child was charged a fair price. If the child had been taller, arrived 5 minutes later, purchased something different, etc., the shopkeeper still would have charged a fair price. On the other hand, certain counterfactuals might exceed those limits, and by considering them we can attempt to clarify the maxim on which he actually acted. If the shopkeeper had been confident that he would be able to charge an unfair price without having any adverse long-term effects on his reputation and business – if he had the mercantile equivalent of the Ring of Gyges – would his deed have been different? The answer to this question would reveal (or at least would provide evidence for) the end and the incentive of his action. Different ends and different incentives mean that the behaviour would have been different under different counterfactuals, and different maxims precisely register these differences.

A maxim is an explicit statement of the causal law that is implicit in an action – ‘maxims [proceed] from choice’ (MM, 6: 226, p.21). When we attribute a maxim to an agent, we need not be claiming that the maxim was present in the agent’s consciousness or anything at all about the deliberative process that the agent engaged in. On the other hand, it is crucial that the agent is capable of reflecting on her actions, exercising practical reason, and determining the maxims on which she acts. These capacities make an agent responsible in general for her maxims whenever she acts, whether she engages in explicit deliberation or not, and whether she is self-consciously aware of her maxims or not. Every action has a maxim because we understand every action to be an instance of causal responsibility, not because anything in particular is explicitly present in consciousness.

Suppose, however, that our shopkeeper did deliberate, engaging his practical reason and surveying the various possible reasons to charge various amounts. In his deliberation, he aimed to identify and evaluate the salient properties of his situation for determining his action. And suppose that while he noticed (or at least hoped) that charging a fair price would help his business in the long run, he was particularly struck by the fact that taking advantage of the child would be wrong, and he recognised that it still would have been wrong if the child had been taller, had arrived 5 minutes later, or had purchased something different. And feeling satisfied with himself, he charged the fair price. Despite his (and our) assumption that he has the capacity to exercise practical reason in determining his maxims, and despite the morally praiseworthy maxim being present in his consciousness, Kant consistently insists that neither he nor we can be certain of the maxim on which he acted: ‘Indeed, even a human being’s inner experience of himself does not allow him so to fathom the depths of his heart as to be able to attain, through self-observation, an entirely reliable cognition of the basis of the maxims which he professes, and of their purity and stability’ (R, 6: 63, p.82; cf. MM, 6: 392, p.166, 6: 447, p.211, G, 4: 407, p.22). Maxims implicate counterfactuals and since we cannot observe these and we cannot be sure what we (or others) would do in counterfactual circumstances, we can never be certain of the maxim on which we act. This is not to say that we can never have evidence regarding the maxim on which someone acts. In the case of natural science, the finitude of our experience will be compatible with indefinitely many laws, and therefore certainty ‘can never be expected’ (CPR, A480/B508, p.505). The same is true with maxims. However, in both cases, empirical observation may provide evidence without rising to the level of certainty. For example, if we observe the merchant charging a fair price in a number of diverse circumstances, especially those in which he believes his reputation would not be affected, and if we believe his sincere avowals concerning his prior deliberations and reflections, we become more confident in our attribution to him of the moral maxim.

4. Multiple maxims

I noted above that by stipulating the maxims in his examples, Kant may obscure the epistemic difficulties we face when attributing them to agents. The examples may also be misleading because the maxims are considered in isolation from others. In fact, the vast majority of our actions (arguably all) are extended over time.Footnote ¹⁴ This means quite literally that we are always doing many things at once, and so there are always multiple maxims in play. This is clearly the case when we consider maxims that describe not specific, concrete actions, but are, in the words of Thomas Hill, Jr., ‘very general, life-governing policies that guide, more or less effectively, our choice of more specific maxims’.Footnote ¹⁵ For example, Kant refers to the ‘maxim of complying with [a] wide duty’ such as beneficence (MM, 6: 390, p.163). For Kant, ‘imperfect’ ethical duties or duties of virtue, in contrast to duties of right, are of ‘wide obligation’ in that ‘the [moral] law cannot specify precisely in what way one is to act and how much one is to do by the action for an end that is also a duty’ (MM, 6: 390, p.163). If one is beneficent and adopts ‘the maxim of making [the happiness (well-being) of others] one’s end’ (MM, 6: 452, p.217), then at appropriate times, one will act to benefit others for its own sake. And then, in addition to the more general maxims, there will also be more particular maxims regarding those specific actions:

Specific beneficent actions will partially constitute being a beneficent person and thereby over time amount to acting on the broader maxim of beneficence. On each of those specific occasions, one will have a subordinate maxim with a particular end that one is striving to achieve, and the incentive will be precisely that the deed will (permissibly) benefit someone else. But of course, in addition to having a general maxim of beneficence and its various subordinate maxims, we will also have indefinitely many other maxims that specify actions related to the goals of advancing one’s career, being a good parent, learning a musical instrument, maintaining one’s health, etc. And although such higher-level maxims may often be compatible with one another – being beneficent and advancing one’s career do not always conflict – this also raises the possibility of practical conflicts among them in specific circumstances.

This complex, hierarchical structure of maxims exists without even introducing considerations of morality. Kant gives slightly different definitions of happiness in different places, but typical is the idea that in happiness ‘all inclinations unite in one sum’ (G, 4: 399, p.14). Owing to our empirical nature, he says happiness is our ‘natural end’ (G, 4: 430, p.42; cf. MM, 6: 386, p.160). Kant notes two important points about this end. First, because it is based on our empirical inclinations – ‘Only experience can teach what brings us joy’ (MM, 6: 215, p.11) – the concept of happiness is ‘indeterminate’: ‘even though every human being wishes to achieve it, yet he can never say determinately and in agreement with himself what he actually wishes and wants’ (G, 4: 418, p.31). Second, our inclinations often conflict with one another. Therefore, we must ‘curb’ some inclinations ‘so that they will not wear each other out but will instead be harmonised into a whole called happiness. Now the reason that accomplishes this is called prudence’ (R, 6: 58, p.78; cf. MM, 6: 385, pp.159-60; L-Eth, 27: 361, p.138). Of course, as always, this process of harmonisation need not be done explicitly, but we do this through the maxims that we implicitly adopt. This system of prudence includes both specific maxims based on our occurrent empirical inclinations and more general maxims, up to and including ‘the maxim of happiness’ (TP, 8: 287, p.288, my emphasis). It is worth noting that Kant does not treat these inclinations or the maxims that are based on them as in themselves objectionable: ‘Considered in themselves natural inclinations are good, i.e. not reprehensible, and to want to extirpate them would not only be futile but harmful and blameworthy as well’ (R, 6: 58, p.78; cf. R, 6: 34-6, pp.57-9; CPrR, 5: 93, pp.78-9). The problems only emerge when those maxims conflict with others, and especially, of course, when they conflict with maxims of morality.

When there is a practical conflict among maxims, agents cannot act on both maxims at the same time. If they subordinate one maxim to another, they will have, in effect, given themselves a more general or higher-order maxim specifying a revised system of ends and incentives together with priorities among them. This subsumption of more particular maxims under more general ones is analogous to the process in theoretical reason by which we explain more particular laws of nature and conflicts among them by identifying more general laws that subsume them and discharge their (perhaps implicit) ceteris paribus clauses. Just as we postulate as a regulative ideal an ultimate causal law that discharges all conditions on subordinate laws and ‘with which its unity will be completed’ (CPR, A307/B364, p.392), we can also postulate as a regulative ideal a single, unified maxim that explains a person’s adoption of all of her more particular maxims. Kant calls this one’s ‘supreme maxim’ (R, 6: 31, p.55) or one’s basic disposition (Gesinnung). He says it is ‘the first subjective ground of the adoption of the maxims, [for which there] can only be a single one, and it applies to the entire use of freedom universally’ (R, 6: 25, p.50).Footnote ¹⁶ Because it is the postulated subjective ground of all of our other maxims, it may or may not be in accordance with the requirements of the categorical imperative. And although Kant sometimes writes as if this maxim is adopted in a single all-encompassing choice, it is more properly understood as being constituted through the totality of our actions over our complete life.Footnote ¹⁷

5. Hierarchy, the strength of maxims, and virtue

Let us consider the shopkeeper one final time and again suppose (contrary to Kant’s assumption) that he was moved to charge a fair price for its own sake because of his recognition that doing so was right. This can be seen as an instance of a more general maxim to act in ways that are just for their own sake. The shopkeeper, being human, also and at the same time has a general maxim of prudence to pursue his own happiness (by satisfying, over time, some constructed combination of his empirical inclinations). This will resolve itself into more particular maxims on particular occasions, and the general maxim to be just will do so as well. In the fortunate circumstances that Kant describes in his first example, these two abstract maxims do not conflict, since charging a fair price will be the right thing to do and will (or may) also make an instrumental contribution to his happiness. But because maxims support counterfactuals, we (and he) can ask what he would do in circumstances in which they do conflict and when an unjust action would better serve the interests of his business or his happiness more generally. Of course, we (and he) may never know the answer unless and until he is forced to make that choice, but if we imagine such a circumstance, we can recognise that through his choice of action he would be subordinating one maxim to the other without necessarily completely abandoning either.

If he is virtuous, he would subordinate his self-interested maxim to his moral maxim when they conflict. Kant says that virtue ‘is the strength of a human being’s maxims in fulfilling his duty’ (MM, 6: 394, p.167). This account introduces a seemingly new idea: the strength of a maxim. Kant is only somewhat helpful when he points out that he is talking about the relative strength of maxims: ‘Strength of any kind can be recognised only by the obstacles it can overcome, and in the case of virtue these obstacles are natural inclinations, which can come into conflict with the human being’s moral resolution’ (MM, 6: 394, p.167). Agents do not passively observe the strength of a maxim. Rather, they give their maxim its strength in comparison to other incentives (as incorporated into other maxims) when they choose among them and act in circumstances of practical conflict. And indeed, Kant gives an alternative definition of virtue that shifts the emphasis from the strength of a maxim to the strength of a will that adopts that maxim: ‘Virtue is … the moral strength of a human being’s will in fulfilling his duty’ (MM, 6: 405, p.175). When an agent’s commitment to one maxim is stronger than her commitment to another, this simply means that the latter is subordinated to the former if and when they conflict.Footnote ¹⁸ But as we have seen, this subordination can be represented by a higher-order maxim that incorporates both, making one conditional on not interfering with the other. Talk of the strength of one maxim compared to another is equivalent to talk about a higher-order maxim that incorporates both and sets priorities between them.

Kant notes that we can, of course, have strong commitments to non-moral or immoral maxims (MM, 6: 384, pp.158-9). He even finds something admirable in such a commitment, regardless of its content:

As we saw above, a maxim pursuing the satisfaction of an inclination is not in itself objectionable. However, our inclinations are inconstant and to some extent unpredictable simply because they are empirical. Therefore, maxims based on inclinations will frequently interfere with our successfully acting on other maxims, especially those that require more complex and temporally extended behaviour, such as those associated with duties of virtue as well as our own overall happiness. I have argued that all maxims are modally robust because they state causal laws. However, higher-order maxims typically require many complexly related subordinate actions and so must be temporally robust as well, and this can be difficult to achieve.

But it may not be clear exactly why it is difficult to act on such higher-order maxims and why Kant would treat this as an accomplishment, especially if we think of giving ourselves a maxim as a purely intellectual act that occurs at a particular instant. Similarly, it may be puzzling why Kant holds that the strength of a maxim could be increased through practice:

Part of the answer, I believe, is that the strength of our commitment to at least our more general maxims, including those associated with wide duties of virtue, only emerges over time since the actions that instantiate them are extended over time, and each occasion that might instantiate the maxim calls on our judgement.

Consider someone who sincerely wants to be a virtuous person and deliberately and self-consciously adopts a maxim of beneficence. Yet, suppose that in situations where a beneficent act would be appropriate, she fails to take the opportunity, either because she does not recognise the situation for what it is or because she has some other end (e.g., one based on inclination) that conflicts with the beneficent act. Since there are very few specific actions that are absolutely required in order to be a beneficent person, when considered in isolation, no single action is inconsistent with her more general maxim of beneficence. But her overall pattern of behaviour is. At the extreme, if she never acted beneficently, we would say she simply has not adopted that maxim, whatever she might say or believe. Short of this extreme, we might say that her commitment to the maxim of beneficence is weak or frail precisely because she subordinates it to many other maxims when they conflict in particular circumstances.

We might explain the weakness of her commitment to the maxim of beneficence as a failure of (perhaps implicit) judgement. As we have seen, ‘ethics … unavoidably leads to questions that call upon judgment to decide how a maxim is to be applied in particular cases’ (MM, 6: 411, p.180).Footnote ¹⁹ If a person’s judgement concerning what a maxim requires in concrete circumstances is sufficiently distorted sufficiently frequently, her commitment to that maxim is weak. A person with seriously compromised judgement may literally not know what she is doing or the maxim on which she is acting. There are many factors that could distort one’s judgement, ranging from alcohol and addiction to various forms of self-deception and rationalisation (in the pejorative sense) based on inclination and self-interest. Kant observes ‘a certain perfidy on the part of the human heart (dolus malus) in deceiving itself as regards its own good or evil dispositions … by which we throw dust in our own eyes and which hinders the establishment in us of a genuine moral disposition’ (R, 6: 38, p.60). Importantly, Kant holds that judgement can be improved through hard work and experience. As he notes in the preface to the Groundwork, although the moral law is itself a priori, it

Experience can help us to apply more general maxims accurately by bringing to our attention salient features of a situation.Footnote ²⁰ For example, Kant holds that we have ‘a duty not to avoid the places where the poor who lack the most basic necessities are to be found but rather to seek them out’ in order to cultivate our sympathy and ability to recognise and share in others’ feelings. (MM, 6: 457, p.221). By cultivating our sympathy, we can become more attuned to what an abstract maxim of beneficence requires more concretely.

Even more important for Kant than nurturing sympathy in this way is the ‘First Command of All Duties to Oneself’: ‘“know (scrutinize, fathom) yourself,” … in terms of your moral perfection in relation to your duty’ (MM, 6: 441, p.206). This command has both a practical and a theoretical component. I have stressed throughout that although the capacity to give oneself maxims deliberately and self-consciously is the ground for attribution of responsibility, we are not always aware – and are never certain – of the maxim on which we actually have acted, even when we have engaged in careful and explicit deliberation. Indeed, explicit practical deliberation is the exception not the rule for most of our actions, and it is, of course, possible to act morally without explicit deliberation. But Kant thinks that a commitment to abstract maxims of virtue that can only be realised over time by many specific acts will be unlikely to survive the onslaught of inclination unless we engage in explicit, self-critical reflection and deliberation. Without this effort, without ‘considered, firm, and continually purified principles’, our commitment to abstract moral principles will be ‘neither armed for all situations nor adequately secured against the changes that new temptations could bring about’ (MM, 6: 383-4, p.158). The same is true of the abstract non-moral maxims concerned with our long-term happiness. That is why Kant finds something admirable in a strong character as such, steadfast in its commitments despite conflicting and inconstant empirical inclinations. And that is why although every action has a maxim, he views acting on such a higher-order or more general maxim, realisable only over an extended period of time, as a genuine achievement of character.

The command to know oneself requires more than taking up the practical, deliberative point of view in order to secure our commitment to these more general maxims. It also demands that we identify and scrutinise the maxims on which we have acted. As noted above, self-rationalisation or a failure of judgement can lead to acting on maxims that diverge from those that we might endorse in our practical deliberations. By attempting to adopt an objective and reflective theoretical perspective on our own conduct, and by considering what we would have done in various counterfactual circumstances, we can gain a more accurate account of our own maxims and thereby test the strength of our commitment to the results of our practical deliberations. This reflective scrutiny of our own actions can be painful when we recognise ourselves to be less admirable than we would like to believe. Yet, this is a crucial component of strengthening our commitment to and more accurately realising our more general maxims of both virtue and happiness: ‘Only the descent into the hell of self-cognition can pave the way to godliness’ (MM, 6: 441, p.206).