The Error Condition

Abstract

The possibility of error conditions the possibility of normative principles. I argue that extant interpretations of this condition undermine the possibility of normative principles for our action because they implicitly treat error as a perfection of an action. I then explain how a constitutivist metaphysics of capacities explains why error is an imperfection of an action. Finally, I describe and defend the interpretation of the error condition which follows.

1. Success and error

1.a

What is the relationship between the nature of our agency and the standard for its exercise? “Constitutivism,” part of a venerable tradition which arguably includes Aristotle and Kant, says that the former determines the latter because what something is determines what it should be, at least with respect to things by nature subject to standards.¹ Yet, despite this pedigree, fundamental doubts about the view persist. In this essay, I consider whether it can explain how the possibility of error conditions normative principles for our agency.

Such a condition differentiates normative from descriptive principles. “Deviance” from a descriptive principle “falsifies” it, as fruit falling too fast falsifies a formulation of the law of gravity. Not so with respect to a normative principle, as getting my kicks on Route 66 by driving on the left does not falsify American traffic laws. Error is thus possible with respect to any normative principle, as here codified with respect to our agency:

Error Condition: P is a normative principle for my action only if I can err with respect to it.

That is one of two related conditions on normative principles. Here is the other, again as it applies to our agency:

Success Condition: P is a normative principle for my action only if I can succeed with respect to it.

Despite this symmetry, the latter, under the name “ought implies can,” receives the bulk of the attention.² Perhaps what explains this discrepancy is an assumption that we can learn what we need just by investigating it. Everyone knows that its consequent concerns what I can do. Why not similarly construe the consequent of the error condition?

1.b

Douglas Lavin argues in “Practical Reason and the Possibility of Error” that constitutivism is incompatible with that interpretation of the error condition.³ He endorses the constitutivist claim that an “exercise of rational agency … involves a commitment to comply with certain principles” because “the capacity to act just is, in part, the capacity to follow the relevant principle[s]” (Lavin 2004, 453).⁴ Yet he claims that if the error condition concerns what I can do, that capacity also “just is, in part, the capacity to violate [that same] principle. But if it is constitutive of agency both to follow and violate a principle, then we can no longer derive an intelligible commitment simply to follow” (454). Constitutivism is thus incompatible with both consequents of these conditions concerning what I can do. Since the consequent of the success condition obviously concerns what I can do, Lavin rejects the symmetrical interpretation of the error condition.

I shall explain and extend this criticism. I explain this criticism because although Lavin implicitly invokes this metaphysics of capacities on which the nature of a capacity establishes a normative principle for its exercise, he never develops it explicitly. I extend this criticism because he misdiagnoses its significance and consequently defends an interpretation of the error condition which is subject to that very criticism. Let me here explain the interpretations he distinguishes so as to clarify what I mean and how I will proceed.

Lavin first contrasts logical and imperatival interpretations. The logical one “says that an agent is subject to a principle only if there is some kind of action such that if the agent did it she would thereby violate the principle” (Lavin 2004, 426). The imperatival one says that “an agent is subject to a principle only if there is some kind of action such that if the agent did it she would thereby violate the principle and it is possible for the agent to do it” (427). Whereas the imperatival interpretation says that something is a normative principle for my action only if I can deviate from it, the logical one says that something is a normative principle for my action only if a possible agent can deviate from it.⁵

Lavin then distinguishes weak and strong imperatival interpretations. The weak version says that “an agent is subject to a principle only if there is some kind of action such that if the agent did it she would violate the principle and it is possible for the agent to do it, and it doesn’t matter why the agent does it” (Lavin 2004, 436). The strong version says that “an agent is subject to a principle only if there is some kind of action such that if the agent did it she would violate the principle and it is possible for the agent to do it and her doing it would be a genuinely practical error” (436). The difference comes out through contrast. Say that I can only commit “error[s] in action that derive from some defect in theoretical reason” (435–36). The possibility of such an agent is compatible with the weak version but not the strong one. If something is instead a principle for my action only if I can commit a basic practical error with respect to it, the strong version is true.⁶

Here are those interpretations:

Strong Imperatival Interpretation: P is a normative principle for my action only if I can commit a basic error with respect to it.

Weak Imperatival Intepretation: P is a normative principle for my action only if I can commit a derivative error with respect to it.

Logical Interpretation: P is a normative principle for my action only if a possible agent can commit an error with respect to it.

The strong version entails the weak one, and both entail the logical one. If I can commit a basic practical error with respect to a principle, I can commit a derivative one with respect to it, at least so long as what is basic in one case can derive from an erroneous exercise of theoretical reason in another. If I can err, a possible being can. The question, then, is not which interpretation, if any, is true but which, if any, is the strongest true interpretation.

Lavin explicitly criticizes the strong imperatival interpretation, but his criticism, if sound, also refutes the weak one. For one thing, he says that the strong version is incompatible with the possibility of “perfect rationality” which “issu[es] only in correct action” (Lavin 2004, 451). Yet if the possibility of basic errors is incompatible with perfection, so is the possibility of derivative ones. The agent still acts incorrectly. For another, he claims that “we can capture the heart of the intuition,” which supports the strong version with the “practical defect constraint: for every kind of practical thought there is a unique kind of practical defect connected with it. It is a further … step to claim that each bearer … must potentially exhibit such a defect” (437n18). If that constraint threatens the strong version, it threatens the weak one too. Both say that each bearer can exhibit a characteristic defect in action. They just differ on the details of that defect.

Since what unites the imperatival interpretations and distinguishes them from the logical one is that their consequents concern what I can do whereas its consequent concerns what a possible agent can do, Lavin locates the source of the problem in that difference. I disagree. I will argue that the mistake is to think that the consequent concerns what anyone can do. Specifically, I shall argue that the metaphysics of capacities on which the nature of a capacity establishes a normative principle for its exercise provides the material for the problem and its solution. I will develop this metaphysics and use it to present his criticism, my extension of it, and my reply to it. The source of the problem is that extant interpretations treat deviance as an activity. They treat it that way because they treat anything whose possibility depends on possessing a capacity as a perfection of its exercise. Yet I shall show that the metaphysics of capacities explains how the possibility of error comes with possessing a capacity even though deviance is an imperfection of its exercise and thus is not an activity in the relevant sense.

The consequent of the error condition thus differs from the consequent of the success condition. Whereas the latter concerns what I can do, the former does not concern doing at all. What it does concern and why, though, will make sense only after I present the metaphysics of capacities. I turn to that task now.

2. Constitutivist capacities

I here explain the aspects of the metaphysics of capacities needed in order to understand Lavin’s criticism of the imperatival interpretations. These are (a) the relationship between a capacity and its state of development and exercise and (b) the relationship between my capacity to act and my capacities to act in various ways. I delay discussing other aspects of this metaphysics until I need them to respond to this criticism.

2.a

Some of what happens in the history of an organism are its activities. These range from the nutritional activities of plants and animals to the volitional activities of animals, including the self-conscious exercises of our capacity to act. As I use the term, capacities are potentialities whose exercises are activities of their bearers whereas vulnerabilities are potentialities whose manifestations are not.⁷ Only organisms can be active and have capacities, and nothing is active with respect to everything in its history. For example, I have the capacity to walk but not the capacity to be blown to bits by the bomb even though I can explode. To walk is to be active. To explode is not. My potentiality to walk is a capacity, my potentiality to explode a vulnerability. A rock, in contrast, is inactive in every event of its history. Hence, whereas I have capacities and vulnerabilities, a rock only has vulnerabilities.⁸ More generally, while organisms have capacities and vulnerabilities, nonorganic things and stuff have only vulnerabilities.

With this active/passive distinction comes another concerning whether a manifestation of a potentiality is as such subject to a normative principle. I walk well or badly, perfectly or imperfectly. I do not explode well or badly, perfectly or imperfectly. Exercises of capacities are by nature subject to normative principles. Manifestations of vulnerabilities are not.

An exercise of a capacity is by nature subject to a normative principle in that the nature of the capacity determines the content of that principle. Compare an exercise of my capacity to add to one of my capacity to divide. These capacities differ given how adding differs from dividing. An account of the nature of a capacity, then, is just an account of the nature of the activity in question. In my terminology, for each capacity there is a principle that describes its nature whose content is identical to the proper description of a certain activity.⁹ And that principle is normative for exercises of that capacity. If I am exercising my capacity to add with respect to 36 and 18, to conclude 54 is correct, 2 incorrect. If I am exercising my capacity to divide, to conclude 2 is correct, 54 incorrect. Which capacity I am exercising determines the principle for the exercise. To exercise the capacity correctly is thus to succeed in doing what the capacity is a potentiality to do. To exercise it incorrectly is to fail in doing what it is a potentiality to do. In my terminology, the principle which describes the nature of a capacity is thereby normative for its exercise.

The nature of a capacity also establishes a principle for its development. That is to say that the principle which describes the nature of a capacity is thereby normative for its development. That is why change is either development or degradation depending on whether it constitutes becoming better or worse at what I have a capacity to do. To develop my capacity to add is to get better at computing sums. For that capacity to degrade is for me to get worse at computing sums. For it to be imperfectly developed is for me to be able to compute some but not all sums. For it to be perfectly developed is for me to be able to compute all sums. Capacities thus can develop and degrade over time and be in a good or bad state. Vulnerabilities cannot. There is no way in which I should be able to manifest my vulnerability to explode. Changes are thus neither development nor degradation. Vulnerabilities thereby lack states of development.

“Capacity” thus picks out a species of potentiality whose members have states of development and exercises that are by nature subject to normative principles. “Vulnerability” picks out another whose members lack those things. Whereas the principle which describes the nature of a capacity is thereby normative for its development and exercise, the principle which describes the nature of a vulnerability is thereby descriptive of its manifestation. Put otherwise, the state of development of a capacity and the exercise of a capacity can be good or bad, complete or incomplete, perfect or imperfect as such. The possession of a vulnerability and the manifestation of a disposition cannot. They just are.

2.b

The principle of a capacity distinguishes it from others by specifying an activity. My capacity to walk differs from my capacity to jump given the differences between walking and jumping. Each, though, is a determination of my capacity to act. Walking and jumping are ways of acting.

To understand this determinable/determination relationship, think about my capacity to speak. I develop it by learning to speak at least one language and exercise it by speaking a language. The only way to develop and exercise the determinable is to develop and exercise a determination of it. No one can simply speak. My capacities to speak, walk, and so on are likewise determinations of my capacity to act. I likewise can only develop and exercise it by developing and exercising its determinations. No one can simply act.

This relationship between my capacity to act and its determinations matters because the determinable restricts its determinations. Take my capacity to walk forward, itself a determination of my capacity to walk. If to walk forward is, in part, to put one foot in front of the other while maintaining balance, the principle of any of its determinations includes this aspect. This principle can modify the first one, but it must stay true to the determinable. That is why strutting, slinking, and sauntering are ways of walking forward but walking while remaining still or walking forward backward are not. Such activities are internally contradictory and hence not real activities. We cannot have capacities to walk in these ways. To set off to so walk is to automatically fail in the way that to set off to square the circle using only a compass and straightedge is to automatically fail. Given the nature of a capacity, then, only certain determinations of it are possible.

The principle of our capacity to act likewise restricts its determinations. For example, since I cannot succeed in action without taking sufficient means to my end, my capacity to act is, in part, a potentiality to take sufficient means to ends. So also are its determinations. We thus cannot have a capacity to sail to the edge of the earth. The earth has no edge and so taking sufficient means to that end is impossible. To set off to so act is to automatically fail in the way that to set off to absolve the sins of a snake is to automatically fail. Taking sufficient means is thereby a normative principle for every human action. Generally, any exercise of any determination of our capacity to act is subject to a version of any rule which is part of the principle which describes the nature of that determinable capacity and is thereby normative for its exercises. Put otherwise, the principle which describes our capacity to act also partially describes all of its determinations and is thereby normative for their exercises.

3. Deviance as doing

Lavin takes as his interlocutor Christine Korsgaard because she is “unique in advancing considerations on … behalf” of the interpretation of the error condition that “lurks in the background of much … work on practical reason” (Lavin 2004, 427). According to him, the mark of this interpretation is that it treats complying with and deviating from principles as on a par. She suggests as much when she claims that “imperatives are addressed to beings who may follow them or not… . [I]t must be possible for a rational being … to disobey, resist, or fail to follow [any] principle” (Korsgaard 1997, 48). Although the second half of this passage is about deviance, the first implies a parallel claim about compliance. Complying and deviating thus seem like distinct activities. So, Lavin reads Korsgaard.

I will ignore interpretative questions. I shall develop a version of Lavin’s criticism which shows that if complying and deviating are distinct activities, our actions are subject to a principle that says either “comply with or deviate from principles” or “comply with and deviate from principles.” The first has the form “do or do not do A,” which violates the error condition and undermines the possibility of normative principles for action. The second has the form “do and do not do A,” which violates the success condition and undermines the possibility of normative principles for action. I will only then ask about which of the heretofore distinguished interpretations of the error condition imply that complying and deviating are distinct activities.

3.a

Consider the relationship between complying and deviating if they are distinct activities. In one respect, they relate to each other like standing and sitting do. Doing one rules out doing the other. In another respect, they do not relate to each other as sitting and standing do. The capacities to sit and stand are different determinations of our capacity to act that stay true to its principle. You can take sufficient means to either end, descending or rising as is fit. Complying and deviating are different. To comply with a principle, I must do what it requires and refrain from doing what it prohibits. To deviate, I must do what it prohibits and refrain from doing what it requires. Deviating does not stay true to any principle. A capacity to deviate is thereby not a determination of any capacity. Since complying mirrors deviating if both are activities, a capacity to comply likewise is not a determination of any capacity.

If the possibility of a normative principle depends on the possibility of complying and deviating, these activities have some relationship to our capacity to act. What relationship? There are two options. Either I have distinct capacities to comply and deviate under whose authority I exercise my capacity to act or my capacity to act is a potentiality, in part, to comply and, in part, to deviate. Both options are incompatible with the possibility of normative principles. I take them up in turn.

If I have distinct capacities to comply and deviate, I only exercise them in an executive role with respect to another capacity. I cannot simply comply or deviate. I must comply with or deviate from some principle. Moreover, I only exercise other capacities under the auspice of one of them. If the principle of my capacity to act says to take sufficient means to my end and I stick with it, I am exercising that capacity under the auspice of my capacity to comply. If I do not, I am exercising it under the auspice of my capacity to deviate. Hence, if I have distinct capacities to comply and deviate, the principle of my capacity to act does not establish a normative principle for its exercise. It does not determine what it is to exercise that capacity correctly. Correctness instead depends on which of my capacities to comply or deviate is in charge.

Whether there are any normative principles for my action then depends on whether the question “Am I to exercise my capacity to comply or my capacity to deviate?” has an answer. It does not. Take any proffered answer. Am I to comply with or deviate from it? Without an answer, the original question is open, and yet this one is a version of the original. Because one capacity says to comply, the other to deviate, every answer in effect says “comply with or deviate from A.” That has the form “do or do not do A,” which violates the error condition and thereby undermines the possibility of normative principles for action. Hence, if I have distinct capacities to comply and deviate, there are no such principles.¹⁰

What if instead my capacity to act is a potentiality in part to comply and in part to deviate? Since complying and deviating conflict in that what one prescribes the other proscribes, acting is then a contradictory activity and hence not a real one. I cannot have a capacity to act any more than I can have a capacity to enumerate the members of the set of all sets who are not members of themselves. It would be a capacity that establishes the normative principle “comply and deviate” for its exercises. That principle has the form “do and do not do A,” which violates the success condition and undermines the possibility of normative principles for action. Hence, if my capacity to act is a potentiality in part to comply and in part to deviate, there are no such principles.

To sum up, if complying and deviating are distinct activities, either I have distinct capacities to comply and deviate under whose auspices I exercise my capacity to act or complying and deviating are both part of the principle of that capacity. The first option implies that every exercise of that capacity is subject to a principle with the form “do or do not do A.” The second implies that every exercise is subject to a principle with the form “do and do not do A.” Since normative principles cannot have those forms, such principles are impossible if complying and deviating are distinct activities.

3.b

That argument refutes any interpretation of the error condition which treats deviance as an activity. Lavin takes this criticism to apply to the imperatival interpretations but not the logical one. That criticism, though, undermines all of them.

Start with the imperatival interpretations. The difference between them is that the strong one invokes basic practical error, the weak one derivative practical error. That difference does not show that only the strong version implies that complying and deviating are distinct activities. After all, the weak version concerns “error in action that derives from some defect in theoretical reason (Lavin 2004, 435–36).” It thus invokes basic theoretical error. Yet, if the possibility of basic practical error implies that deviating and complying are distinct activities, so does the possibility of basic theoretical error. Such error is still deviation from a principle. Hence, if the strong version is subject to this criticism, so is the weak version.

If those interpretations imply that complying and deviating are distinct activities, so does the logical one. What distinguishes them from it is whether something is a normative principle for my action only if I or a possible agent can deviate from it. This difference does not change the type of error invoked. The logical interpretation either invokes derivative practical error and thus basic theoretical error, as in the weak imperatival interpretation, or basic practical error, as in the strong one. After all, if it were to invoke another type of error, there would be a third imperatival interpretation which would say that I am subject to a principle only if I can commit that type of error with respect to it. Hence, there are two logical interpretations as there are two imperatival interpretations:

Strong Logical Interpretation: P is a normative principle for my action only if a possible agent can commit a basic error with respect to it.

Weak Logical Interpretation: P is a normative principle for my action only if a possible agent can commit a derivative error with respect to it.

If the imperatival interpretations imply that complying and deviating are distinct activities, then so do these logical interpretations. They all treat error in the same way. The imperatival ones concern what I can do, the logical ones what a possible agent can do, but all concern what someone can do. All treat deviance as doing, as an activity. To use Lavin’s language, they all invoke a “kind of action such that if the agent did it she would thereby violate the principle” (Lavin 2004, 426; my emphasis). That is the problem. It has nothing to do with what differentiates these interpretations from each other.

Hence, the consequent of any true interpretation of the error condition does not concern what anyone can do. Whether there is such an interpretation, though, does not turn on whether its consequent concerns me or a possible agent or whether it concerns derivative or basic practical error. The problem and solution instead turn on the account of error.

4. What I can do and what can happen

I here introduce an aspect of the metaphysics of capacities which explains why complying and deviating are not distinct activities. I explain the interpretation of the error condition which follows in the next section.

4.a

The relevant aspect of the metaphysics is a distinction between classes of what is possible given possession of a capacity. There is what I can do given that I possess the capacity and what can happen given that I possess it. These classes are a package deal but are not on a par because the latter depends on the former. Let me explain.

If the principle which describes the nature of a capacity is thereby normative for its exercise and if a normative principle allows for success and error, possible exercises of a capacity divide into the successful and the erroneous. Likewise, possible properties of exercises divide into perfections, which contribute to success, and imperfections, which contribute to error. Take my capacity to walk forward. Because I have developed it, I can put one foot in front of the other while maintaining balance. In this way, a capacity makes possible the perfections of its exercises. It also makes possible characteristic imperfections of its exercises. If I can walk, I can trip, slip, stumble, or tumble. Any being who can walk, but only a being who can walk, can so err. Neither a rock nor a goldfish can walk or trip. Walking is doing what I have a capacity to do. Each step is a perfection which contributes to the success of an exercise. Tripping is an imperfection which contributes to the failure of that exercise if it is severe enough. A walk to the store is successful though imperfect if I stumble along the way. A trip which ends only halfway to the store with me flat on my face with a fractured ego and an even worse leg is an erroneous exercise of that capacity.

Whereas walking is something that I can do given my capacity to walk, tripping is something that can happen given that capacity.¹¹ My capacity to walk thereby brings with it two subclasses of what is possible with respect to it. This terminology is a bit awkward. “What can happen” might sound like it only picks out things that come from outside the exercise of the capacity. That impression is misleading. Imperfections can come from outside the exercise, as when a tornado ends a walk. Imperfections can also be internal aspects of the exercise, as when a false step and the fall and resulting fracture end a walk. There are thus internal imperfections, which happen in the exercise of the capacity and express a defect in the state of development of the capacity, and external ones, which happen in the exercise but need not express a defect in the state of development. Less grandly, when I err, sometimes I screw up and sometimes something else messes me up. Some errors are my fault. Others are not.

Although these subclasses of what is possible with respect to a capacity are a package deal, they are not on par with each other. A capacity is a potentiality to do something, and the nature of a capacity is tied to the nature of the activity in question. What differentiates my capacity to walk from my capacity to sing is, after all, the difference between walking and singing. Tripping and the like are possible imperfections in exercises of my capacity to walk because of the nature of walking. They are imperfections because they get in the way of success in a way that burping does not. When singing, though, burping matters. The possible perfections and the possibility of success of an exercise of a capacity in this way precede its possible imperfections and the possibility of error. What can happen given a capacity depends on what I can do given it because what can happen are interferences with, interruptions in, and other imperfections of what I can do.¹²

Although my terminology here is unique, others mark this distinction with different terminology. Immanuel Kant says that “freedom in relation to the internal law-giving of reason is really an ability; the possibility of deviating from it is only an inability” (Kant 1797, 6:227).¹³ In my terminology, complying with a law of reason is what I can do, deviating what can happen, given that I possess practical reason. Gilbert Ryle likewise says that

if a person can spell or calculate, it must also be possible for him to misspell and miscalculate; but the sense of “can” in “can spell” and “can calculate” is quite different from its sense in “can misspell” and “can miscalculate.” The one is a competence, the other is not another competence but a liability. (Ryle 1949, 131)

In my terminology, spelling and calculating are what I can do, misspelling and miscalculating what can happen, given my capacities to spell and calculate.

4.b

As the possibilities of walking and tripping are with respect to our capacity to walk a package deal but are not on a par, so are the possibilities of complying and deviating with respect to our capacity to act. In fact, walking and tripping are specific determinations of complying and deviating. To walk is to comply, to trip to deviate, in an exercise of my capacity to walk. To comply, generally, is thus to do what I have a capacity to do. To deviate, generally, is to err in doing what I have a capacity to do. Complying is a perfection and deviating is an imperfection in an exercise of our capacity to act. If the imperfection is severe enough, I erroneously exercise my capacity and fail in that exercise.

Complying and deviating are thus not distinct activities. Only complying characterizes the nature of our capacities and thus only it is an activity. Complying is what I can do, deviating what can happen, given my capacity to act. The possibility of deviating thus depends on the possibility of complying.

While it does not matter for my argument, Lavin and Korsgaard should agree with this account of error. Consider that Lavin claims that “the idea of a capacity or power to resist reason is a confusion on the order of the idea of a capacity not to see, or the treatment of blindness as itself a capacity” (2004, 450n48). Just as susceptibility to blindness, whether temporary or permanent, comes with the capacity to see, so susceptibility to error comes with our capacity to act. Still, blindness is not on a par with seeing but is derivative of it. Deviance is likewise not on a par with compliance but is derivative of it. Consider also that Korsgaard’s prime example of error is “something … interfering with … reason” such as when I am “rendered inert by depression, or paralyzed by terror, or … simply unable to face” acting some way “because the means are painful” (1997, 49). These things are imperfections, not perfections. Aside from her calling them “interferences,” look at the verbs: I am rendered inert or paralyzed, not deliberately standing still; I am unable to face something, not deliberately looking away. Both of them, then, should accept that complying and deviating are a package deal but are not on a par because complying is what I can do and deviating is what can happen given my capacity to act.

5. The possibility of error and the possibility of perfection

That account of error is compatible with altered versions of the earlier interpretations of the error condition. Just change their consequents so as to concern what can happen given what someone can do rather than what anyone can do. In other words, make them treat error as an imperfection rather than as a perfection.

Altered Strong Imperatival Interpretation: P is a normative principle for my action only if a basic error with respect to it can happen in my action.

Altered Weak Imperatival Interpretation: P is a normative principle for my action only if a derivative error with respect to it can happen in my action.

Altered Strong Logical Interpretation: P is a normative principle for my action only if a basic error with respect to it can happen in the action of a possible agent.

Altered Weak Logical Interpretation: P is a normative principle for my action only if a derivative error with respect to it can happen in the action of a possible agent.

Again, certain interpretations entail others. The strong imperatival interpretation entails the others. If a basic error can happen in my action, a derivative one can, at least as long as what is basic in one case can derive from an erroneous exercise of theoretical reason in another. If those errors can happen in my action, they can happen in the action of a possible agent. The weak imperatival interpretation entails the weak logical one. If a derivative error can happen in my action, it can happen in the action of a possible agent. The strong logical interpretation entails the weak one. If a basic error can happen in the action of a possible agent, a derivative one can, at least as long as what is basic in one case can derive from an erroneous exercise of theoretical reason in another. The question is thus again not which interpretation, if any, is true but which, if any, is the strongest true interpretation. There are two questions to answer. Is it strong or weak? Is it imperatival or logical? I address them in turn.

5.a

Is the consequent of the error condition about basic or derivative errors? Lavin defines derivative practical error as error in an exercise of our capacity to act which derives from an erroneous exercise of another capacity such as theoretical reason. That definition grounds a negative definition of basic practical error as error in an exercise of our capacity to act which does not so derive. What are examples of such errors? Philosophers tend to focus on dramatic examples, like Satan declaring “Evil, be thou my good” (Milton 1667, IV.110). These examples tempt us to treat deviance as an activity. Satan does not simply fall from grace but instead takes up arms against it.¹⁴ Focusing on mundane cases of error, though, tempers that temptation.

When I am on the ground with a fractured leg and ego, I fail in my exercise of my capacity to walk. This failure might derive from an erroneous exercise of theoretical reason. Maybe I miscalculated the terrain. It might not, though. I might just take a false step. Such a step is an error, a bit of deviance from the principle of my capacity to walk, which does not derive from an erroneous exercise of another capacity. Or maybe “the order is ‘Right turn’ and I turn left: no doubt the sergeant will insinuate that my attention was distracted, or that I cannot distinguish my right from my left—but it was not and I can, this was a simple, pure mistake” (Austin 1956a, 148n1). Anyone can so err. Such errors are basic practical errors whose possibility comes with having a capacity to walk or a capacity to turn right on command. Yet they do not imply that deviating is an activity, that tripping or turning right are perfections of the exercises of the capacities in question. Such errors and imperfections are legion. They include mistake, accident, inadvertence, incompetence, inattention, neglect, clumsiness, carelessness, and much else besides. Whatever the differences between them, they are all imperfections of exercises of my capacity to act, not perfections. They can happen given determinations of that capacity because of what I can do given those determinations.

The possibility of some of these imperfections comes with the possibility of any determination of our capacity to act. The possibility of deviance generally likewise comes with the possibility of our determinable capacity to act. After all, each of these specific types of deviance is a way in which I might not take sufficient means to my end in the exercise of a determination of my capacity to act. Perhaps I do not take sufficient means because of an erroneous exercise of theoretical reason. I might be wrong about how to pull off the end. Perhaps, though, I know but still mess up. The possibility of such basic errors comes with possessing a capacity to act. They are imperfections in exercises of that capacity, not perfections, and not doings in the sense that implies that complying and deviating are distinct activities. Hence, the consequent of the error condition concerns basic practical errors.

Such an account of basic practical error and the resulting interpretation of the error condition “treat[s] deviations from the relevant standard as we might treat, say, blindness in a human being, [as] an accident, and not as an expression of what rational agency is” (Lavin 2004, 454). After all, having the capacity to see brings with it the possibility of blindness, and not every instance of every type of failure to see derives from an erroneous exercise of another capacity. For all that, the capacity to see is still a potentiality to see. Blindness is not on a par with sight even though the possibility of one is tied to the possibility of the other and even though the source of the blindness might not be an erroneous exercise of another capacity. Just so, having the capacity to act brings with it the possibility of deviance, and not every instance of every type of error derives from an erroneous exercise of another capacity. For all that, the capacity to act is still in part a potentiality to comply. Deviance is not on a par with compliance even though the possibility of one is tied to the possibility of the other and even though the source of the deviance might not be an erroneous exercise of another capacity. Hence, the strongest true interpretation of the error condition is one of the strong interpretations.

5.b

Is the consequent of the error condition about me or a possible agent? Lavin has a second argument against the imperatival interpretations. He claims that they are incompatible with the possibility of “perfect rationality” understood “as a state of will nonaccidentally issuing only in correct action, though only contingently possessed by its bearer” (Lavin 2004, 451). Such perfection is possible if the success condition is true. If the imperatival interpretations rule out that possibility, the strongest true interpretation of the error condition is the strong logical one.

Why might the imperatival interpretations be incompatible with perfection? Lavin’s argument has the following premises:

• 1. If perfect rational agency is possible, an agent can have a perfectly developed capacity to act.

• 2. If an agent can have a perfectly developed capacity to act, her exercises of it are nonaccidentally correct.

• 3. If those exercises are nonaccidentally correct, error is not a live possibility for her.

• 4. If an imperatival interpretation is true, error is a live possibility for every agent.

The possibility of perfection would then entail the falsity of the imperatival interpretations. This argument, though, is unsound. The third premise is false. Lavin asserts it because he thinks that if “there is some circumstance in which X, such as he is, would go wrong, and if it is only an accident that X does not find himself there, then for any state of will that X is in, it must be an accident that X does not go wrong” (Lavin 2004, 451). The key to understanding why this claim is false is to first distinguish my capacity to act as I by nature possess it from my state of development of it, and then distinguish internal and external imperfections in the exercise of a capacity. I take up these tasks in turn.

A capacity is a potentiality to do something. A being with a perfectly developed capacity is as skilled as possible at that activity. If I perfectly develop my capacity to act, I always exercise it as and when I should in virtue of my knowledge of what I am to do, when I am to do it, how I am to do it, and so on. Hence, if I fully develop this capacity, I always pull off what I am to do, at least absent outside interference, and I do so because I have so developed this capacity. The perfection of the state of development of the capacity thus explains the success of its exercise. Success is no accident.

Because this capacity develops, I possess this state of development contingently, and yet it leads to success in its exercise through its own perfection. That meets Lavin’s description of perfect practical rationality as a state of will nonaccidentally issuing only in correct action, though only contingently possessed by its bearer. Is it incompatible with the live possibility of error? No. That possibility is, in fact, part of this “terrestrial … conception of a perfectly rational agent” (Lavin 2004, 450). Even if I fully develop my capacity, I might fail in some exercises of it because of forces outside of my control. For example, say I am exercising my capacity to build a monumental sandcastle. I head to the beach and start designing, digging, dragging, and developing. While I am acting, the action is incomplete, ongoing, and to that extent indeterminate. I have made the walls, the bastions, and the turrets. The steeple and the palace and all that jazz are still to come. My acting gets more determinate as I act and, while I act, I know what I am doing and why given the state of development of my capacity to build sandcastles. Despite my talent, though, I might not pull off the action. Even if I build further back than the water has ever reached, an eruption of a previously undiscovered underwater volcano might cause an unprecedented wave to wash out the castle before I finish it. I might end up mangled floating in the puddle of mud which sits where the castle once rose, left with nothing but dreams of the majesty of what I will never complete. I fail in that exercise through no fault of my own. All the skill and planning and foresight possible might not be enough to stop the weather or other wills or what have you from wiping me out. I will then fail in my exercise of my capacity to act even if I fully develop it and thus am perfectly practically rational. After all, this capacity is the fully developed human capacity to act, and some of the world is beyond the ken and control of even the greatest among us. It can disrupt me in ways that I cannot predict and which I cannot halt, avoid, or overcome.

But, honestly, so what? An account of our agency is not about a being whose capacities do not develop or degrade and who knows and controls everything that can affect success in action. Such a being could not have a capacity or any other kind of potentiality, let alone a state of development of it, but would be pure actuality, whatever that might mean, and infinite. That is not the type of perfection needed in a metaphysics of capacities of organisms because it “is not in line with the traditional beliefs enshrined in the word can: according to them, a human ability or power or capacity is inherently liable not to produce success, on occasion, and that for no reason (or are bad luck and bad form sometimes reasons?)” (Austin 1956b, 166n1; original emphasis). Why? Because I am finite and depend on what lies outside me. My capacities thereby come with the possibility of compliance and deviance, of success and error, because they are capacities of a human being.

Some, but only some, errors are possible even if I fully develop my capacity to act. If I so develop that capacity, I am able to exercise it in all the ways that I by nature should, and I exercise it in all of those ways when I should. There are thus no internal imperfections in the exercise. No false steps or bad form. Hence, if I succeed in the exercise, my success exemplifies virtue and is no accident. That success, though, still depends on the world around me. Even a walk to the store in which every step is internally perfect succeeds only if no sinkholes open under me, no snipers pick me off, Zeus does not zap me, and so on for the myriad of ways that the world can mess me up. Hence, even a perfectly rational human being can err and fail in the exercise of our capacity to act in virtue of external imperfections which derive from our dependence on the rest of the world.

That dependence on the world is not an imposition on our exercises of our capacity that makes a successful exercise accidental, as if my capacity to act is in itself detached from my finitude. The world and the other organisms in it are not directly subject to my will. My capacity to act is thereby a potentiality to act in a world among others like and unlike me which is partially but not wholly within my ken and partially but not wholly within my control. Dependence on the world is part of the nature of that capacity and part of the nature of success in exercising it. Acting without depending on the world is a contradictory human activity and thus not a real activity at all. We cannot have a capacity to act that way anymore than we can have a capacity to breathe something other than oxygen. The fact that the success of the exercise of our capacity to act depends on the world does not make that success accidental any more than the fact that the success of the exercise of our capacity to breathe depends on the presence of oxygen makes that success accidental. Without this dependence, there is no such capacity and no possibility of exercising it. To insist otherwise is to confuse perfect human rationality with infinite agency, to reject an “ideal actually attainable in the life of that which operates with it” for an alien alternative (Lavin 2004, 450).¹⁵

That dependence on the world explains the possibility of errors which do not indict perfection. Lavin is right to insist that perfect rational agency is possible for me only if it is possible that the success of my exercise of my capacity to act is “no accident” and expresses my perfection of that capacity. He is wrong to think that the live possibility of error is incompatible with this perfection, though, because he does not distinguish between types of errors. This perfection is not only compatible with but in fact depends on the live possibility of certain types of errors because it is the perfection of a capacity of a finite being. Perfection is incompatible with the possibility of internal imperfections in the exercise of my capacity to act. They exemplify defects in the state of development of that capacity and thus imperfection. It is compatible with the possibility of external imperfections. They need not exemplify defects in the state of development of that capacity, or of any other for that matter, and thus need not exemplify imperfection. Perfect rational human agency is compatible with the live possibility of error in action and, in fact, with the actuality of error in action, because sometimes external imperfections are severe enough for me to fail.¹⁶

Perfection is thus possible for us, and we therein advance beyond the obstacles within. The obstacles without, though, are not so easily overcome. Even if some are for a bit and others are forever, there is no escaping them all. Error is possible because I cannot know and control everything that can affect my success. Austin is thus half right and half wrong. The fully developed human capacity is not liable to error because of bad form but is so liable because of bad luck. Korsgaard is likewise half right and half wrong. She is right to say that “there is no normativity if you cannot go wrong” (Korsgaard 1996, 161). She is wrong to say that “a perfectly rational will cannot be conceived as guided by reason unless it is conceived as capable of resisting reason,” at least so long as “resisting” exemplifies an internal imperfection which expresses a defect in the state of development of the capacity (Korsgaard 1997, 52n39). A perfectly rational will can be guided by reason because it must take means to ends in a world of obstacles and opponents which might disrupt it and undermine its exercise without indicting its perfection. The race is not always to the swift.

The consequent of the error condition thereby concerns me, not a possible agent, because human beings are organisms and thus finite. Basic practical errors can happen in the exercise of the capacity to act of even a perfect human being. This fallibility makes success in action for even the perfectly rational agent contingent in the sense that it depends on the world. Changes in the world might result in failures without corresponding changes in the state of development of the capacity to act. It does not make success accidental, though, because such dependence is part of the nature of the capacity itself. It is simply the way it is for a being like us in a world like ours. Error and failure are possible for every human being. Hence, the proper interpretation of the error condition is the altered strong imperatival interpretation.¹⁷

6. A human being in the world

Something is a normative principle for my action only if a basic practical error with respect to it can happen in my action. Korsgaard is right that the possibilities of success and error are a package deal, but she errs when she lets this insight lead her to imply that complying and deviating are on a par. Lavin is right that complying and deviating are not on a par, but he errs when he lets this insight lead him to imply that the possibilities of success and error are not a package deal. You cannot do right if you cannot go wrong, cannot go wrong if you cannot do right. The metaphysics of capacities explains why in three parts.

• 1. The lesson of Lavin’s criticism is that although complying and deviating are a package deal, they are not on a par. Deviance is what can happen, not what I can do, given my capacity to act. It is an imperfection which, if severe enough, constitutes a failure in an exercise of our capacity to act. Hence, the consequent of the error condition concerns what can happen, not what I can do, given my capacity to act.

• 2. Since the principle which describes my capacity to act is normative for its exercise, properties of exercises divide into perfections, which contribute to the success of those exercises, and imperfections, which contribute to the failure of those exercises. Such imperfections might derive from the erroneous exercise of another capacity. They might not, though, but might instead be internal or external basic practical errors. Hence, the consequent of the error condition concerns basic practical errors, not derivative ones.

• 3. The possibility of a perfectly rational human being is compatible with the possibility of basic practical errors. Such a being possesses a fully developed capacity to act. If I fully develop that capacity, I do not err because of internal imperfections which express a defect in the state of development of the capacity. My successful exercises are thus no accident. External imperfections in the exercise of the capacity, though, are still possible. A human being is finite, and a mark of finitude is that the world is not wholly within our control or ken. I might then err because the world, or others in it, undermine my exercise. Hence, the consequent of the error condition concerns basic practical errors which can happen in my action.

This interpretation of the error condition partially explains what it is like to be agents like us in a world like ours. As human beings, we habituate ourselves and each other to virtue in thought and action in the face of obstacles from within and without. So the metaphysics of capacities says insofar as the principle which describes the nature of our capacity to act is thereby normative for its development and exercise. Perfect practical rationality is the perfection of the capacity to act of an organism whose attempts to understand, navigate, and shape the world can faultlessly fall short. It is the perfection of the capacity of an organism who lives in a world which can elude attempts to understand and alter it. It is the perfection of the capacity of an organism who lives in a world among other beings who can undermine attempts to understand and act with them. That might seem like a concession, but it is not. It is just what it is to be a human being. As Kant says, “Human beings can err: the ground of this fallibility is to be found in the finitude of human nature” (Kant 1763, 2:202).¹⁸

Acknowledgments.

Thanks to James Bondarchuk, Matthew Boyle, Colin Chamberlain, Ursula Coope, Edward Hall, Christine Korsgaard, Douglas Lavin, Patricia Marechal, Richard Moran, and Thomas Pendlebury. Thanks also to audiences at Harvard University, Temple University, and the 2018 Bechtel Workshop at Northwestern University and to the referees and editors for this essay.

Jeremy David Fix is a fixed-term tutor in philosophy at Keble College and a departmental lecturer in the faculty of philosophy at the University of Oxford.

¹ I explain the basic metaphysics of constitutivism in “Two Sorts of Constitutivism” (Fix, forthcoming). Although this view of human agency is a specific determination of that metaphysics, I here stick to claims about our agency for the sake of simplicity.

² I use the name “Success Condition” in order to suggest the link between these conditions.

³ That essay is the only focused discussion of the error condition. See (Cokelet 2008) for the only reply.

⁴ See (Lavin 2017) for his defense of a version of Aristotelian constitutivism.

⁵ Lavin formulates the logical interpretation in terms of a “type of action such that if the agent did it she would thereby violate the principle”, but he does not specify what counts as a “type of action”. This threatens to make the logical interpretation vacuous. For example, “A and not A” might count as a type of action for all he says. My formulation fends off this threat by linking the possibility of a type of action with the possibility of a type of agent who can so act.

⁶ Lavin does not positively define basic practical errors. I shall later.

⁷ Something like this notion of a capacity pops up in recent Kantian and Aristotelian practical philosophy. See, among others (Quinn 1992, 210), (Thompson 2008, part 3), (McDowell 2010, 5), (Elizondo 2013, 4–5), (Lavin 2017), and (Shafer forthcoming). Although these authors tend not to thematize the contrast with other types of potentiality, the distinction is often implicit in which potentialities they do not discuss. Regardless, the examples I present justify the distinctions. No one denies the difference between walking and exploding or between an animal who is sometimes active and a rock which is not. Terminological regimentation in terms of capacities and vulnerabilities just minimizes confusion. To test whether a potentiality is a capacity or a vulnerability of an organism, just ask whether the organism can develop or exercise that potentiality well or badly, perfectly or imperfectly. If so, it is a capacity. If not, a vulnerability.

⁸ “Capacity” here encompasses what a venerable tradition calls active and passive powers. My capacity to hear is a passive power, my capacity to walk active. Both types of powers differ from my vulnerabilities given that their states of development and exercises are by nature subject to normative principles. “Potentiality” picks out the genus of which capacities and vulnerabilities are species; “capacity” picks out the genus of which active and passive powers are species. To say that a capacity is a potentiality to do something is thus a bit misleading because I am the patient, not the agent, of the causal transaction when I exercise a passive power.

⁹ Talk of “description” is a bit misleading. “Add” is a fine “description” even if sometimes I need or want to say more. Such further “descriptions” elucidate rather than explicate. They are not analyses of the first one.

¹⁰ The same argument works against a view which says I have a single capacity to comply or deviate.

¹¹ Perfections and imperfections are relative to the capacity in question. Whereas maintaining balance is a perfection, tripping an imperfection, of the exercise of my capacity to walk, I might exercise a different capacity, say, one to take a dive, which reverses the perfections and imperfections. This is why “the clown’s trippings and tumblings are the workings of his mind, for they are his jokes; but the visibly similar trippings and tumblings of a clumsy man are not the workings of that man’s mind” (Ryle 1949, 33). I ignore this relativity here, though, because it does not matter for the claim that what can happen given that I possess a capacity depends on what I can do given that I possess it.

¹² In addition to errors in exercises of a capacity, I can also fail to exercise a capacity. The metaphysics thus must explain how a capacity brings with it standards for when to exercise it. I ignore this aspect of capacities here, though, because it does not matter for interpreting the error condition.

¹³ Kant asks, “How can the former be defined by the latter?” and says that such a definition “puts the concept in a false light” (Kant 1797, 6:227). Although he does not say so, this passage implies that the inability is defined in terms of the ability. This implication is clearer in the original since “ Unvermögen,” here translated as “inability,” has connotations of failure or deterioration in eighteenth-century German.

¹⁴ Although I am indulging this reading, I disagree with it. The surrounding passages imply that this declaration expresses despair and desolation because of Satan’s fall from grace and the consequent impossibility of reconciliation with God. “Evil, be thou my good” does not express the clear-eyed conviction of an enemy. It expresses the resigned recognition of irredeemable rot wrought by the wretched.

¹⁵ Lavin elswhere claims that although “central features of human agency are incompatible with divine perfection,” he does not “want to deny the intelligibility of the idea of divine agency. Nevertheless, whatever intelligibility it has, it has through other channels than giving application to the forms of predication at the ineliminable core of our own intentional agency” (Lavin 2013, 296n3). Likewise, if divine agency is intelligible, we understand it through channels other than giving application to forms of capacity predication at the ineliminable core of our own intentional agency.

¹⁶ A full discussion of error would distinguish two types of imperfect states of development. Some errors express an incomplete state of development, as when a doctor kills through lack of skill, others a perverted state of development, as when a doctor kills through the malicious use of skill. This difference does not matter in this essay, though, because a perfectly rational human being will not err in either way.

¹⁷ There is a question about the way to understand intentionally deviating from principles within a constitutivist account of human agency. While important, this question is irrelevant to the discussion of the error condition. The possibility of error holds even for the virtuous human being, who does not intentionally deviate because it is an internal imperfection which expresses a defect in the state of the development of the capacity to act. The proper account of intentional deviation, then, presupposes the more basic account of error in the error condition and must be understood against that background.

¹⁸ David Walford renders “ Der Mensch kann fehlen ” as “human beings are capable of error.” “Human beings can err” is truer to the German, though, and consistent with my terminology into the bargain.