The Use of Reason An Essay on Practical Wisdom

In his Nicomachean Ethics, Aristotle makes what might strike us as an outrageous

claim. He says that studying ethics is only worthwhile for those who are already good

(2002, 1095a3-13). What might seem even more outrageous is that part of the ground

for this remark is the idea that ethics is characteristically practical, meaning that its

point lies not merely in the knowledge of truth, but in action. Wouldn’t the practicality

of ethics indicate that it is especially those who are not already good who should study

it? If those who study it are already good, then what is left for ethics to do?

Aristotle’s thought, in answer to the first question, is that studying ethics will

have no effect on anyone who is not already basically good. Let her study all she wants

—if someone is not motivated to do as reason commends, what would argument

change? Argument is, after all, addressed to reason.

The argument bears the stamp of Aristotle’s elitism and might therefore arouse

our suspicions. Is it not too quick, not to mention unduly pessimistic? Yet, while it

receives little immediate elaboration, there is no question that the argument has some

basis in Aristotle’s considered moral psychology, on which our ends are set by the

characters we acquire during our upbringings (cf. Moss 2011). That is not, of course, to

say that it is beyond question, but it does mean, perhaps, that it is due some

consideration.

Two sorts of answers might be given to the second question, about the point of

studying ethics for those who are already good. The one that seems to me to have the

clearest textual basis is that Aristotle’s Ethics is meant to be continuous with his

Politics (2002, 1094a22-b12; 1180a1525). Ethics is practical, at least, in the sense that

it can help already virtuous rulers arrange their cities in such a way as to produce

virtuous citizens and to encourage virtuous activity. To do that well, they need to have

more than a passing idea of what virtuous activity is and what are the conditions of its

possibility.

That answer is perhaps not so satisfying to those who hope for the study of

ethics to make a difference to their own lives, who expect it to help them in answering

that illustrious question: what am I to do? A different sort of answer might interest

them, however. Namely, although the student must be basically good if ethics is to have

any effect on her, there may remain some nontrivial course correction that ethics can

achieve. Ethics, perhaps, can unearth principles at work implicitly in the thinking of the

basically virtuous and help them to apply such principles more consistently, or to hard

cases.

Such a conception of the practicality of ethics, on which it articulates principles

that are serviceable at least for those whose hearts are in the right place, is not

altogether without promise. It seems not, though, to be what Aristotle means to

accomplish in his ethical works. He is not very concerned with the explication of

principles or with their application to hard cases, after all. He even, famously, suggests

that general principles about what to do hold only “for the most part” (2002, 1094b20-

23).

I dwell on this not to foreclose the possibility that ethics can be practical in such

a way. Aristotle is not the final word on ethics. Nevertheless, the question about the

practicality of ethics raised by Aristotle’s paradoxical remark seems to me to be

important and interesting. From Aristotle, I wish to take the idea that a carefully

articulated moral psychology can call into question some of the aspirations of ethical

theorizing—or, to put the point more positively, can help us appreciate how necessary

for the achievement of ethics’ purpose are the theoretically untutored dispositions of

ordinary virtuous people. Aristotle’s argument exploits what he takes to be a conative

requirement on ethics’ making a difference in our lives: ethical argument will be

pointless if directed towards those whose characters do not set for them the right ends.

My contribution will be to exploit a cognitive requirement on ethics’ making a

difference: one can only act well if one is able, on the spot, to determine what acting

well is. This disposition is, I suppose unfortunately, not something which the study of

ethics imparts.

An example might help to illustrate this theme. Rosalind Hursthouse recounts a story

which G. E. M. Anscombe once told in a lecture:

An old woman in Austria under Nazi rule had given shelter to some

Jews in her attic, and one evening, there was the dreaded knock on the

door and a young SS officer saying “We believe you have some Jews

here?” “Clearly,” said Anscombe, “she must not lie.” And there was a

long embarrassed silence because we all thought that obviously she

must—that this was the “morally right” thing to do—but did not dare to

say so. And she let the pause continue, and then said “Of course, she

mustn’t tell the truth either,” and we were all greatly relieved, but also

puzzled.

She went on to describe what the woman had in fact done; she had

turned on a brilliant performance of pretending to believe that the young

officer was her sister’s son, whom she had not seen since he was a boy.

“Gustav!” she cried, “how wonderful, come in, come in. How is dear

Lotte, I haven’t heard from her for so long, I never knew you had

become an officer, how tall you have grown … !” And she kissed him

and babbled on (never once telling a lie) and insisted he have coffee and

cakes and, being young and well-mannered, he was too embarrassed to

tell her she had made a mistake and press his official question. So he

partook of the coffee and cakes and escaped as soon as he could. (2008,

142-143, quoted in Richter 2011, 21-22)

Of course, one central aspect of this example will be controversial. Many of us are not

absolutists about lying, and if there were anyone to whom we would lie without a

second thought, it would be the proverbial Nazi at the door. One might also query

whether something like the old woman’s performance really is better than a simple lie.

So set those features of the case aside, if you like. Most of us will still agree that we are

not to lie with impunity, even where the person inquiring is not entitled to the truth. An

inquirer who is not entitled to the truth is still, often, entitled not to be lied to.

Suppose, then, that someone has asked you what she reasonably thinks is an

innocent question about some matter which happens to be private. There is a

confidence you must not betray, but you also should not lie. I think many of us will

agree that such cases exist. They are a regular part of navigating our complicated social

world. But the question is: in the circumstances, can you see how to do this—how to

avoid betraying the confidence and telling the lie? That may be no easy matter. And yet

you may be faulted for fumbling in the situation, for coming out quickly with a lie, or

else betraying the confidence unintentionally, by (for instance) your hesitation. Maybe

the right thing to do is to lie, but you conclude that too late for your lie to be effective.

In any case, what you don’t manage to figure out is how to do what needs to be done,

whatever that happens to be—at least, you don’t figure that out in time for the doing of

it. What you lack is what we can call, following Aristotle, practical wisdom, or simply:

wisdom. The topic of this dissertation is what it is to be wise.

My approach is governed by an old Aristotelian principle, according to which

powers— such as that of human reason—are to be understood through their activities,

since after all powers

are simply capacities for those activities. Accordingly, I begin, in Chapter 1, with a

treatment of reasoning. I distinguish between two related acts which are sometimes

called “reasoning”: on the one hand, the activity of one’s reasoning, the sort of thing

that takes time and sometimes effort; on the other, one’s rationale for, say, thinking or

doing something, the sort of thing on the basis of which one thinks or acts, the sort of

thing one produces in response to the questions “Why do you think that?” and “Why

are you doing that?”. Chapter 1 introduces this distinction and urges its importance in

underwriting an account of what it is to reason (in the activity sense) well. I leave the

topic of the standard of goodness in rationales to Chapter 4.

What the distinction makes plain, I argue, is that reasoning well is not simply a

matter of conforming to rules of reasoning (for instance, rules of inference), however

abstractly they might be conceived. It importantly must involve, for example,

remembering and noticing things which turn out to be helpful for the solving of

whatever problem or question one is confronted with. And so, to be a good reasoner,

one must be such that one tends so to remember and notice; one needs a particular kind

of disposition—a disposition, to put it briefly, for good mental conduct, which might

not unreasonably be called rationality, or in the more general practical case, wisdom.

In this way, my consideration of the nature of reasoning approaches the topic of

virtue, which is after all, as Aristotle says, a disposition. What sort of thing a virtue is is

the topic of Chapter 2. There I seek to extract and defend a conception of virtue which I

find in the later work of Philippa Foot, the goal being to articulate an account of virtue

in which the dispositional account of rationality developed in Chapter 1 can neatly be

situated.

I argue that the role of human nature in Foot’s argument for the rationality of acting

well is not what it has seemed to be to many philosophers. Foot has been taken by the

readers I have in mind to hold that facts about human nature are among the premises of

(good) practical reasoning. They have understandably balked at that thesis, which for

various reasons seems so implausible. I argue that Foot’s basic argument for the

rationality of acting well does not rely on any such claim. This becomes evident if one

gives appropriate priority to Foot’s conception of virtue as a certain kind of disposition,

a disposition to respond to what I call situational reasons—particular facts about one’s

situation such as that one made a promise or that someone else is in need. The account

of reasoning and rationality developed in Chapter 1 fits cleanly into this account of the

nature of virtue, for rationality as conceived there is a sine qua non of virtue in Foot’s

sense.

While Chapter 2 aims to rescue Foot from some of her readers, Chapter 3 aims

to rescue Foot from herself. Foot’s good account of the rationality of virtue is, I think,

entrenched within an objectionable account of the teleology of life. I raise two kinds of

objection to Foot’s account of life. First, I think it renders puzzling just how virtue

could be a disposition to recognize and respond to the right sorts of situational reasons,

because such a disposition must also, given Foot’s neoAristotelian naturalism, be a

disposition to recognize certain truths about human nature. Foot’s conception of human

nature in terms of human need—what she calls Aristotelian necessity— renders such a

disposition incredible, or so I claim. The second objection concerns a kind of vicious

circle in Foot’s account. Foot is pulled into the circle by the particularities of her

attempt to extend her account of life and natural goodness to human beings, in which

attempt she tries to satisfy desiderata that pull in opposite directions. I argue that an

account of life which shelves the notion of Aristotelian necessity can respect those

desiderata without being subject to the vicious circle.

Importantly, the account also makes space for a sensible and realistic epistemology of

morals and in turn of human nature.

Chapter 4 completes the account of wisdom by providing an account of what makes

for a good—valid, sound—practical inference (or rationale), since good practical

inferences are what, after all, it is the business of practical wisdom to make. I treat first

the instrumental component of practical inference, arguing that it is best conceived as

identifying sufficient rather than necessary means to one’s end and that it should not be

modeled too closely on theoretical inference. Second, I argue that situational inference

—inference from situational reasons to actions—is also a fundamental component of

(at least many) practical inferences. Taking note of this fact helps us to formulate a

neo-Aristotelian account of the interdependence of wisdom and the various virtues of

character and likewise to give accounts of practical reason and the will.

Chapter 1: Reasoning and Rationale

Deliberation is a certain sort of inquiry. (Aristotle 2002, 1142a32)

1.1 Introduction

In one sense, reasoning is something that we do. It is a process that develops over time.

It is an activity, to some extent under our control. In another sense, we sometimes say

to someone, for instance, “Explain your reasoning.” In this sense, reasoning is a

particular kind of ground for belief or action, an account that one normally would or

could give if asked. Finally, by ‘reasoning’ we might mean a certain kind of abstract

object, namely an argument or proof.

I will reserve the term reasoning for reasoning in the first of these senses, and I will

refer to someone’s reasoning in the second sense as a rationale. Though it is not my

aim to treat the third, I will assume that we all have a serviceable sense of what is

meant by ‘argument’: a collection of considerations, one of which (the conclusion) is

meant to be grounded or supported by the others (premises).

The topic of this chapter is the nature and assessment of reasoning. The goal is

to give an account of reasoning in general, of what is common to both theoretical and

practical reasoning. In doing so, I will develop some conceptual resources that I will

make use of in the remainder of my argument.

The argument will proceed by developing and defending the distinction

between reasoning and rationale, a distinction which seems to me important and

underappreciated. I will argue that reasoning is a kind of inquiry, the aim of which is a

rationale, or a response on the basis of a good rationale. Seeing this distinction aright

sheds some light on, and accommodates well our pretheoretical conception of, what it

is for someone to reason well. Negatively, I will suggest that the failure to appreciate

the distinction between reasoning and rationale has led to distortion of the assessment

of reasoning. Positively, I want to urge that in a correct account of good reasoning,

essential reference must be made to the concept of a disposition. Good reasoning is,

very roughly, the sort of reasoning a good reasoner is liable to engage in.

1.2 Activity and Argument-Evaluability

In recent literature, there are two features reasoning has commonly been supposed to

bear.

One recurring idea is exemplified by the emphatic remark that reasoning is

something we do. As I will put the idea:

(1) Reasoning is active.

By this what is meant is not that ‘reason’ appears in English as an active verb. ‘Believe’

appears in English as an active verb, but in a natural, intuitive, and familiar sense,

believing is not something we do. The thought meant to be captured here is that

reasoning involves agency.

The expression of this point varies from author to author. Some call reasoning a

process (Harman 1986; Wedgwood 2006; Boghossian 2012; Broome 2013; Dancy

2018). Some add that it is an action or activity, a voluntary action, or an intentional

action (Boghossian 2012; Wright 2012; Broome 2013, 207, 234ff; Ford 2016, 155-

156). Most agree that reasoning is in some degree under the control of reasoners. The

idea raises questions that are not always answered. Reasoning will be casually

described as being an attempt or as having an aim (Boghossian 2012, 5; Dancy

2018, 6), without an apparent consequence of this—that reasoning is an intentional

action— informing the account, or else being explicitly addressed and denied. There

are various disagreements in matters of detail, concerning, for instance, the extent to

which reasoning involves effort or is deliberate. I conjecture that these disagreements

are in large part traceable to differences in how authors stipulate their subject matter—

whether, for instance, they mean to treat reasoning in general or inference more

specifically, and what sort of examples they make central. There does seem, though, to

be a signal in this noise, a truth which various authors are approaching: simply, that

there is such a thing as active reasoning, reasoning as an activity or “something we do.”

Reasoning in this sense is what we are talking about when we say of some subject that

she is reasoning, that reasoning is what she is doing.

Active reasoning is a process. It bears, that is, the characteristic marks of

processes: it takes time, has a describable trajectory, can be interrupted, and can be

ascribed to subjects in progressive aspect. More than that, it is indeed an intentional

action; those whom we can describe as reasoning are able to say why they are

reasoning, what goal they are seeking by engaging in this activity. This would explain

why it is natural to say, as many authors do, that reasoning has an aim, takes effort, or

is a kind of attempt. The claim that reasoning is active distinguishes it from such

attitudes as belief and intention, for believing and intending cannot be interrupted, only

bear progressive aspect in extended or ironic senses, are not aptly described as

attempts, and so on.

The second recurring idea I want to highlight is this:

(2) Reasoning is argument-evaluable.

According to (2), reasoning is evaluable according to the standards of arguments. What

I mean by this is most evident in accounts of inference itself, for inference involves the

conformity of a reasoner’s mind to certain canons of logic (Boghossian 2012, Wright

2012, Broome 2012). Even philosophers whose topic is reasoning in some broader

sense tend to agree that inferences are instances of reasoning, and indeed they tend to

treat them as paradigmatic instances. In the development of a theory of reasoning, it is

usually a modus ponens inference that is wheeled out, at least initially, as an example

of the phenomenon in question. What routinely follows is the acknowledgment that

reasoning need not take the form of simple schematic inferences; reasoning may be

abductive or practical, for instance, and it is generally messy, enthymematic, or

inexplicit.

Even with such qualifications in place, active reasoning is commonly

understood in the following way. It is taken to proceed from premises to conclusion,

such that the subject’s endorsement of the premises explains her adoption of the

conclusion in virtue of their being its ground, and the adequacy or success of the

reasoning is a matter of whether its premise- and conclusion-attitudes (or their

contents) are appropriately related—a matter of whether the premises really do ground

or justify the conclusion (Kolodny 2005, 520; Wedgwood 2006; Broome 2013, 235ff).

In other words, philosophers often take active reasoning to be a process of an

argument’s unfolding in time within the thinking of a subject, and they take it that

reasoning is as such to be assessed. Related ideas include the views that reasoning

proceeds in accordance with rules (Boghossian 2012, 11ff; Broome 2013, 235ff), that

reasoning is a kind of working with or manipulation of propositions (Broome 2013,

234), and that the contents of reasoning are (or should be) actual or perceived reasons

(Harman 1986, 2; Kolodny 2005, 529; Schroeder 2007, 26).

The identification of these two features widely held to characterize reasoning

allows me to frame the first major claim I will be making in this chapter—against (2)

so long as in it ‘reasoning’ refers to active reasoning, as most philosophers writing on

the topic so take it to refer:

(3) Reasoning is not argument-evaluable.

This is not, of course, to deny that there is an activity in which we engage, largely

under our control, which we call reasoning, nor to deny that this activity is central to

the conduct of our lives, to our deciding what to think and what to do. But it is to deny

that that activity is, and is to be evaluated as, the unfolding of an argument in time. To

the extent it can be said to have contents, the contents of active reasoning need be

neither propositions nor reasons. A fortiori they need not constitute an argument. The

trajectory of active reasoning is not from premises to conclusion of arguments,

deductive or otherwise. What is genuinely evaluable according to standards of

argument are what I call rationales, the finding and enacting of which is, roughly, the

aim of the activity of reasoning.

1.3 An Example of Active Reasoning

In this section, I want to present my case for (3). I will not argue for (3) by suggesting

that activity and argument-evaluability are incompatible, that there could not be a

process that is both active and argument-evaluable. My principal argument is rather:

“don’t think, but look!” (Wittgenstein 1958, §66). Philosophers have conceived active

reasoning to be argument-evaluable in large part because in developing their theories

they have nourished themselves upon “a one-sided diet” of examples (Wittgenstein

1958, §593). They attend to and give priority to the unrepresentative and

misleading cases of straightforward deductive and instrumental inferences, supposing

that even if not all reasoning is of one of those sorts, then they at least supply the basic

cases by comparison with which others should be understood. Accordingly, to motivate

(3), I want to present a case which I take to be, and to be worth calling, reasoning, if

anything is. My case indicates the need for a distinction along the lines that I am

urging. Since many philosophers treat reasoning in some restricted sense, this example

is not intended as a direct counterexample to anyone’s views. It is rather meant to

reorganize our conception of our subject matter and remind us of possibilities we easily

overlook.

My example is this: You are a student who has been asked to prove that there are

infinitely many prime numbers. You do various things in order to complete this task; in

the remainder of this section, I will speculate somewhat upon this variety. At the end of

all of this, you have written down, say, the following:

Suppose toward a contradiction that there are only finitely many primes

p1, p2, … , pn. Let A be the integer p1 · p2 · … · pn + 1. A is either prime

or composite. For each i, pi divides p1 · p2 · … · pn, but does not divide

1, so for each i, pi does not divide A. So A is not composite, since it is

greater than 1 and is not the product of primes lesser than it. So it is

prime. But p1, p2, … , pn were assumed to be all of the primes, and A is

greater than each pi. So A is not prime. This is a contradiction.

Therefore there are infinitely many primes.

In a natural and familiar sense, what you have done in coming up with this proof is

mathematical reasoning. About this activity, it seems to me that several observations

can be made.

(a) This reasoning is an activity; it is, intuitively, “something you do.” It is a

process; again, it takes time, has a describable trajectory, could be interrupted, and is

ascribed to you in progressive aspect. So your reasoning is, in my terminology, active.

Accordingly, what goes on in a case such as this seems likely to be of interest to

philosophers whose topic is active reasoning.

(b) Your reasoning is an intentional action. It is done for a purpose, which you

could supply were you asked. Your goal is to prove that there are infinitely many prime

numbers. As is frequently the case in the study of mathematics, you are aware from the

context, perhaps from the testimony of your instructor or textbook, that there are

infinitely many prime numbers, so in this case you are simply seeking a proof. But the

question might alternatively have been put to you whether there are infinitely many

prime numbers, which would be more akin to the sorts of questions practicing

mathematicians actually address, and that would affect the shape of your reasoning in

various ways. Then you would be seeking not simply an answer, for you could get that

by guessing randomly, but for an answer that you can reasonably take to be correct.

You seek a true ground for some answer.

attempt to close an open question, and that is what you attempt to do in this example. I

will generalize from the present example by proposing the following thesis:

(4) Reasoning is a kind of inquiry.

Why think that (4) holds quite generally? I invite the reader to survey a variety of cases

to test its adequacy. It seems that, in general, we will find it necessary to distinguish the

trajectory of reasoning from its argument-evaluable issue. Consider, for instance,

scientific reasoning: what grounded Einstein’s conviction in his theory of relativity was

not an accounting of what he was up to, what he did, in the patent office in Bern. Or

consider instrumental reasoning: one reasons how

to attain some end, starting from a conception of the end but (perhaps) no idea even of

what candidate means are available and sufficient. Non-instrumental practical

reasoning could be said to be governed by the question what to do. Even where

reasoning is not aimed at the evaluation of a determinate thesis, as when, for instance,

one learns a new fact and wonders what follows from it or how it is compatible with

what else one believes, one’s thinking is governed by an open question.

(4) only claims that all reasoning is inquiry, not that all inquiry is reasoning. For

some inquiry seems not to be reasoning. Mere counting, for instance, may be how one

comes to answer a question. Breathing deeply and slowly would not be called a

technique of reasoning, even if it is part of one’s procedure in a hot moment for

deciding what to do. Simply asking someone a question is one form of inquiry, but is

not reasoning. Finally, one might resolve an open question by way of experiment, as

when one manipulates a tool to decide whether it would serve one’s purposes, and this

too is not obviously what we would call reasoning. (Though it also does not seem

obvious to me that we would not call it reasoning.) This raises the question of whether

there is some articulable specific difference which would distinguish reasoning from

the genus of inquiry more broadly—but it is not necessary for my argument to

articulate such a difference, which, as far as I am aware, may not exist, for the

boundaries of our concept of reasoning need not be sharp.

(d) Like inquiries generally, your reasoning proceeds from doubt. If you

were already to know what to do to prove the theorem, then you would not devote so

much time and effort to reasoning it out. This is not to deny that one sometimes

reasons about a question one has already answered. In such cases, one seems to be, as

it were, simulating doubt, or trying to decide whether there is another proof.

Since your reasoning proceeds from doubt, the course your reasoning should

take is not, in general, obvious at the outset. It may not be obvious, for instance, that

the proof that there are infinitely many prime numbers should proceed by

contradiction. Some students will begin by trying other strategies, like mathematical

induction; they might try writing down the first several primes and attempting to find a

general formula for the nth prime in terms of the previous ones. And even those who

start by assuming that there are finitely many prime numbers are faced with a decision

how to proceed from that assumption.

(e) Crucially, because your reasoning proceeds from doubt, it tends to

involve a variety of mental acts besides calculation and the application of rules. Among

this variety of acts are guesses. These may be more or less methodical, more or less

educated. For instance, in proving the theorem, you need to come to consider the

number p1 · p2 · … · pn + 1—but how do you come to consider this? If you are a savvy

mathematician, you may be in some sense guided by the principle that if a divides b

and does not divide c, then it does not divide b + c, and that means both remembering

this principle and noticing its salience in the particular case. If you are not a savvy

mathematician, you may simply guess, knowing only: that you likely need to make use

each of p1, p2, … , and pn; that primeness has more to do with multiplication than with

addition; and, yet, that the product of your primes won’t do, since it is manifestly

composite.

Another sort of mental act your reasoning may involve is imagination.

Imagination seems to play a role in a lot of mental arithmetic. For instance, in this case,

you might try to imagine pi for arbitrary i being factored out of p1 · p2 · … · pn + 1,

yielding an integer plus, you notice, a nonintegral fraction. Such imaginative activity

should be reckoned as part of your reasoning too—for your reasoning’s success may

well depend on it. Imagination can play an obvious role in instrumental reasoning—for

instance, in determining whether a particular material will be suitable as a lever.

Arguably, imagination has a role to play in non-instrumental practical reasoning; one

might try to place oneself in someone else’s shoes, say, to decide whether some

outcome is fair.

Because you are liable not to know how to begin, reasoning may involve

supposition and other sorts of work that do not directly contribute to your ultimate

proof. For instance, the mathematician Terence Tao advises the following when one is

unsure how to proceed in proving a theorem: considering an extreme or degenerate

case; solving a simplified version; and trying first to disprove the theorem in question

(2006, 4-5).

The key idea is that reasoning cannot in general proceed by the application of

rules, however abstractly those rules are to be formulated. As we might put it,

reasoning must proceed in an ad hoc manner.

Of this variety of acts that can be involved in reasoning, what seem to me to be

most significant are memory and the recognition of saliences. Those will be my focus

in what follows.

(f) Your reasoning is susceptible to a form of evaluability besides argument-

evaluability.

As I will put it:

(5) Reasoning is means-end-evaluable.

Reasoning involves taking means to an end and is evaluable on the basis of the

conduciveness of those means to that end—for all reasoning is intentional action,

inquiry in particular.

Reasoning proceeds through various steps. I do not mean to deny that some

steps of some reasoning are argument-evaluable, as would be the case if they were

transitions from ground to what is grounded. For example, if you make the assumption

for contradiction that there are only finitely many primes, then this might be said to be

an argument-evaluable step, for logic permits one to make assumptions from an empty

set of premises.

Even so, there are indefinitely many assumptions which logic permits one to

make at any point, and there are a variety of consequences that can be drawn from any

set of premises. Most of these logically permissible steps would not only be unhelpful

to take, but obviously so, and the reasoner who takes them would be said, in a familiar

sense, to be reasoning badly. This is why acts such as remembering, noticing, and

guessing what is salient are, in real cases, essential to reasoning’s going well.

Recognizing what sort of assumptions (for instance) will be helpful to make in the

course of proving a theorem is what separates good from bad mathematical reasoners.

The idea of taking a step in accordance with rules, however abstractly those rules are

conceived, does not settle the question of which step it would be helpful to take, nor

which step one does take.

Put another way, your reasoning is going badly if you cannot come up with the

strategy of argument by contradiction or if you cannot make use of your assumption by

coming to consider the number p1 · p2 · … · pn + 1. But your mistake is not to have

stepped out of conformity with standards of argument. What in general makes your

reasoning go well is for the steps you take to be conducive, or likely conducive, to

finding the proof.

(g) This point about the conduciveness of means to one’s end needs to be

handled with care. Of course, means-end-evaluations must be sensitive to the nature

and difficulty of one’s question (as the evaluation of any intentional action is sensitive

to the nature and difficulty of attaining one’s goal). The fact that reasoning proceeds

from doubt means that reasoners cannot in general light immediately upon promising

means of solving their problems. Such is the lot of finite reasoners. In the present

example, it need not redound very negatively upon one’s competence at reasoning to

attempt mathematical induction before trying argument by contradiction. Neither

strategy is foolish to consider, even though only one of them in fact works in the given

case. It may redound negatively upon one’s competence to spend too long on an

unfruitful attempt before turning away from it, but what counts as ‘too long’ will

depend upon the nature of the case. In any case, the sensibility of directing one’s

efforts in a particular direction does not reduce to correct application of a rule; even

where guessing is methodical (that is, where one enumerates possibilities according to

some method or rule) the method of enumeration must occur to one.

(h) Conversely, the fact that one’s reasoning does successfully light upon a

correct proof does not mean that it was good reasoning, for it may have so lighted by

accident or luck. Some students recently apprised of the power of proof by

contradiction will try it immediately in proving every theorem with they are

confronted; this means that they will sometimes get an answer more quickly than other

more circumspect students, but they accomplish this, when they do, only by accident or

luck, and in other cases their wheels spin fruitlessly or they produce inelegant proofs

containing unnecessary steps. Their selection of approach was not responsive to the

nature of the problem that confronted them. Yet methodical guessing is not always

possible, and some difficult problems do simply require that one has some luck or tries

sufficiently many possibilities so as to light upon the correct one.

(i) It is true that one way in which reasoning can go badly is calculative

error. One might make a mistake in doing some mental arithmetic, say, by forgetting to

carry the 1. That this sort of mistake is possible is, probably, a great source of the

pressure to consider active reasoning to be argument-evaluable tout court. But all that

need be true in such cases is that one misapply a good technique—with the result that

what one thinks on the basis of such miscalculation is mistaken.

In other words, the steps taken in reasoning are all means-end-evaluable, but

they need not be argument-evaluable. Even those that are argument-evaluable are also,

and more fundamentally, means-end-evaluable. Reasoning can go badly even if one

does not transgress any standards of argument. Indeed, reasoning can go badly even if

one is doing nothing other than applying rules correctly.

I believe that a fair survey of other sorts of reasoning will bear out that these

observations generalize straightforwardly to regions of thought besides the

mathematical. I have chosen a mathematical example to illustrate my theses for a

couple reasons. First, antecedently to our considering a particular case, many of us will

be inclined to think that mathematical reasoning is an especially plausible candidate for

rule-governed reasoning. If anything is a paradigmatic instance of reasoning’s

purported argument-evaluability, one would think, then surely it must be mathematical

reasoning. Yet, second, while it is clear that the product of mathematical reasoning—

the ultimate proof—is argument-evaluable, it becomes equally clear, in the inspection

of an actual case, that the reasoning leading up to that product is not. Reasoning

proceeds in all sorts of ways.

The clarity with which a mathematical example can be described helps us to see this.

1.4 Rationales

None of this is meant to suggest that reasoning is a free for all. The appellation

‘reasoning’ suggests that the process is a rational one. My proposal is that this

rationality is a matter, not directly of how reasoning proceeds, but rather of what it

aims at.

But what does it aim at? The example of the previous section might lead one to

say that reasoning aims at a proof or an argument. That is all right, if it is understood in

the right way. The wrong way to understand it would be if one meant that reasoning

aims at producing a proof in the sense of an abstract object, as a carpenter’s activity

aims at a house—for abstract objects, of course, cannot be produced. Rather, it is the

discovery of an argument or proof that one is after; one wants to believe some answer

to one’s question, on the ground of an argument. To mark this distinction, I say that

what one aims at is a rationale, rather than a proof. In this section, I say more about

what I mean by ‘rationale’.

To get the notion of rationale in which I am now interested in view, I find it helpful to

reflect upon G. E. M. Anscombe’s account of the practical syllogism in Intention

(2000, 57-81, hereinafter cited as I). Anscombe turns to the topic of practical reasoning

or practical syllogism (she says that these are the same) because she thinks it will help

us understand practical knowledge, about which she thinks we ought to be puzzled.

The sections of Intention on practical reasoning contain a battery of interesting and

suggestive theses, most of which do not concern us here, as well as interpretation of

Aristotle. I want to highlight what Anscombe argues a practical syllogism does and

what it does not do.

Whereas a theoretical syllogism shows the truth of a proposition, a practical

syllogism does shows the point of an action (I, 60). What a practical syllogism does not

do is describe “actual mental processes” or considerations that agents “go through”

before or in acting, for often what is made explicit in the representation of a practical

syllogism is not the sort of thing agents “go through” at all. Her examples are such

thoughts as “I am human” and “Lying on a bed is a good way of resting.” She says that

Aristotle’s practical syllogisms have “an absurd appearance” because it is absurd,

implausible, to imagine someone thinking such things before taking a nap (I, 79). But it

is in fact not absurd to include such thoughts in a practical syllogism, because its

function is otherwise than to report how someone has deliberated: “In several places

Aristotle discusses them only to point out what a man may be ignorant of, when he acts

faultily though well-equipped with the relevant general knowledge.” The premises of a

practical syllogism are supposed to represent what an agent knows, or at least believes,

insofar as this sheds light on what she does:

The interest of the account is that it describes an order which is there

whenever actions are done with intentions; the same order as I arrived

at in discussing what ‘the intentional action’ was, when the man was

pumping water [pumping water, that is, in order to replenish the water-

supply of a house, in order to assassinate the party chiefs housed there,

in order to put some good men in power, in order to bring about the

Kingdom of Heaven on earth]. (I, 80; cf. 37ff for Anscombe’s example)

So Anscombe’s view is that the practical syllogism covers the same ground as her

special sense of the question “Why?” by appeal to which she defines the class of

intentional actions (I, 5-28). A theoretical syllogism would correspond to a different

question “Why?” (“Why do you think that?”), the answers to which are reasons for

belief rather than reasons for action (cp. Anscombe, 1981a, 229-230).

This draws out the fact that rationales have two dimensions. As answers to questions

“Why?”, they provide explanations of belief and action. But they also indicate what the

subject takes to support or justify believing or acting as she does, for they show the

truth of some conclusion or the point of doing something. Further, these two

dimensions, explanatory and justificatory, are connected. Rationales provide

appropriate and truthful answers to explanatory questions “Why?” in virtue of

indicating what the subject takes to support or justify her belief or action. In other

words, the distinctive way in which a rationale explains a belief or action is by giving

the ground or justification appreciated by the subject.

This dual role of explanation and justification is familiar from the philosophical

debate over the nature of reasons (e.g., Williams 1979). Rather than speaking only of a

subject’s reasons, I have chosen ‘rationale’ to gesture toward the fact that a subject’s

reasons for a belief or action

are structured into a case or argument for her belief or action. Since we ask whether

this case is a good one independently of the truth of the agent’s reasons, we may say

that rationales are assessable on the basis of something akin to validity: that is, whether

the case really does ground the response. In other words,

(6) Rationales are argument-evaluable.

This accommodates, I want to say, what seeming truth there was in (2).

Now, Anscombe famously gives her account of intentional action in terms of a

piece of language, a certain kind of question. This Wittgensteinian approach is

controversial, and its execution calls for a delicate navigation between Cartesianism

and behaviorism—a major theme of Intention (I, 41-49). My argument in this chapter

does not depend on Anscombe’s method. Even philosophers who do not wish to adopt

Anscombe’s approach should agree with this: there are such things as rationales for

beliefs and actions, and they are not merely abstract arguments, because it is only by

belonging in some sense to a subject that they discharge their explanatory function.

Further, like believing and intending themselves, having a rationale for one’s belief or

action is not a process; to have a rationale for a belief or an action, to believe or act for

reasons, is not to be thinking of something at a particular moment or for a time.

Further, it cannot be held that what it is to believe or act on the basis of a

rationale is itself provided by an account of active reasoning, by an account of the

process of reasoning that leads up to belief or action. For reasoning is itself done on the

basis of some rationale. If you interrupt

someone’s reasoning with the question “Why are you doing that?”, the reasoner will be

able to give an answer that provides a rationale for engaging in some reasoning on this

particular occasion, and this rationale must be different from the one being sought—

since that, of course, has not yet been found. And so if we were to hold that actions are

done for reasons in virtue of issuing from deliberation, we would face a vicious

regress.

Active reasoning is an essential part of how human beings come to know the

world and act within it. But since reasoning is itself an action, acting for a reason

cannot be accounted for exhaustively in terms of reasoning. Our account of acting for a

reason needs rather to appeal to what I have called rationale.14

1.5 Doing Without the Distinction?

The interest of the distinction between reasoning and rationale is not just that it renders

explicit an ambiguity in the ordinary word ‘reasoning’. The distinction is important

because the failure to recognize it, embodied in the tendency of philosophers to treat

active reasoning as argumentevaluable, tends to obscure what it is for reasoning to go

well. In defending my conception of reasoning, I have at times appeal to our sense of

what it is for reasoning to go well, but now I want to consider how neglect of this

distinction tends to lead to obscurity about the assessment of reasoning.

The tendency to see reasoning as argument-evaluable is the tendency to see it as

proceeding from premises to conclusion. John Broome, for instance, takes this to be a

representative example of the topic he aspires to treat:

You believe it is raining and you believe that if it is raining you ought to

take an umbrella. As you head for the door, thinking about what to pick

up, you bring these beliefs together in your mind and end up believing

you ought to take an umbrella.

Going through this process is something you do. (Broome 2013, 207)

Such a picture of reasoning emphasizes the role of logical (or at least quasi-logical)

rules, along which the steps taken in reasoning run. Rules, of course, are normative.

They provide standards to which activity may conform or fail to conform. This

conception of reasoning, accordingly, appears to supply the following picture of what it

is for reasoning to go well rather than badly: reasoning goes well when it actually does

conform to rules of reasoning.

The example is typical of the literature on reasoning and inference. Broome

acknowledges that it is unnatural, for it is difficult, he says, to find good examples of

what he too calls active reasoning. He conjectures that this is because it happens too

fast. “To minimize this problem,” he says, “I shall imagine your mind is slowed down

by dopiness.” This seems to me to be a sign that something has gone wrong. It should

not be difficult to come up with examples of reasoning, for the progressive action

descriptions ‘is reasoning’ and ‘is deliberating’, no doubt, commonly have application.

The mistake is to think that they apply to people who are precisely in the middle of

drawing inferences—indeed, that there is such a thing as being in the middle of

drawing an inference. For, notice, ‘infer’ is a near synonym, in many contexts, of

‘assume’, and like the latter, does not bear progressive aspect and so does not signify a

process that can be interrupted.

But let us set aside Broome. It is instructive to consider in detail Jonathan Dancy’s

account of reasoning. For one might think that if anyone could accommodate the

messiness of actual episodes of reasoning, then it would be Dancy, since he has long

urged the attractions of particularism in ethics (cf. Dancy 1983; 2006), and accordingly

has denied that rules or principles play a preeminent role in reasoning.

Active reasoning, in the sense adumbrated in Section 1.2, is the topic of Dancy’s recent

book Practical Shape (2018). He holds that reasoning is “an attempt to determine what

sort of response is most favoured by the considerations at issue, and responding in that

way if one can” (2018, 6). He represents such reasonings on what he calls grids, the left

column of which name the reasoner’s attitudes, the right their contents, like so:

Belief p

Belief If p then q

Belief q

Table 1: Dancy's modus ponens grid

Although this grid looks like an abstract argument, Dancy cautions that it is meant to

represent rather an episode of reasoning, the reasoning someone does on some

occasion, and such episodes are processes, not arguments:

[T]he grid represents a process, not a steady state. A deliberator adduces

considerations one after another, and comes to act (or believe or intend)

accordingly. So there is a temporal order from top to bottom. Of course

it is not vital that the process should have had this rather than that order,

and often actual cases of reasoning are pretty much of a jumble anyway.

(2018, 25)

Further, even though his treatment begins with a schematic example of modus ponens

reasoning, Dancy means ultimately to be faithful to his particularist roots. Codifiable

rules of inference are not to play any special role in his account of the nature of

reasoning (2018, 54-55). For Dancy, what is essential to reasoning is the assemblage

and appreciation of considerations that favor some response, and considerations might

favor a response even without conformity to any general rule.

This abstract favoring relation thus characterizes how, according to Dancy, reasoning’s

startingattitudes and concluding-attitudes ought to be related, if reasoning is to go well.

So even a particularist such as Dancy, who holds that reasoning in actual cases is

“pretty much of a jumble,” takes reasoning to be argument-evaluable in the sense I

specified in Section 1.2.

Dancy’s question is how the considerations adduced in his grids are (or should be)

related to the response in which they issue. His answer is that they (should) favor it.

But when we consider a realistic example of an episode of reasoning—like the case of

mathematical reasoning of Section 1.3—we can appreciate that this account is not

really an account of what it is for reasoning to go well but rather of what would make a

complete, filled-in grid provide a good rationale for some response.

Dancy acknowledges that if we are to represent someone’s reasoning, then we need

really to understand the grids as possessing a temporal dimension, as being filled in

gradually. But having made this point, he makes nothing of it. He continues to take the

problem of the nature and assessment of reasoning to be a problem of how the

reasoner’s attitudes are to be related to the response, rather than (say) how the reasoner

comes to have the relevant attitudes that ultimately ground that response. For Dancy,

the fact that reasoning is “pretty much of a jumble” does not mean more than that the

considerations which compose his grids might come in any temporal order (and, we

may add, may be inchoate, inexplicit, or enthymematic).

We might put the point this way: Dancy says that reasoning is an attempt to

determine what is the most favored response to the considerations “at issue.” But in

what sense are there considerations at issue? The relevant considerations are either, as

it were, given, or they are not. If

they are assumed to be given, if the reasoner is assumed already to be aware of them

once reasoning is taking place, then the account of reasoning fails to appreciate that

reasoning often begins without any premises. So it was in the mathematical example of

Section 1.3. This is clearly not a special feature of mathematical reasoning—

instrumental reasoning often begins without the agent’s having any sense of, for

instance, what means are available and sufficient for the attaining of her end. If the

relevant considerations are not assumed to be given, then the reasoner’s becoming

aware of them is itself a significant problem which the reasoner faces, the resolution of

which is not itself a matter of determining what response is most favored by other

considerations at issue, on pain of regress. In either case, Dancy’s account assumes that

the success of reasoning hinges on considerations’ favoring, rather than not favoring,

the response. Bad reasoning turns out to be reasoning that possesses, in a highly

abstract sense, a kind of invalidity. It is a transition from grounds to a response which

those grounds do not actually favor.

It is no wonder that Dancy does not find illuminating examples of reasoning in

some of the places where he looks. He writes that he tried to test and develop his theory

by modeling the reasoning found in legal judgments. But the results were

underwhelming:

The written judgments were often impressively long, which looked

promising, but large parts of the judgments usually consisted of an

elaborate laying out of the case at issue, and then a statement of the

judgment itself. The reasoning itself was usually fairly brief—one

paragraph or two at most. I remember one that was of the form

‘We just don’t do things this sort of way in Britain.’ (2018, 57)

But Dancy should not have expected to find what he was looking for here. Legal

judgments do not report on active reasoning—on how a judge reasoned in deciding a

case, on the process a judge went through in figuring out which judgment to render.

They are meant to explain the judgment and justify it to a particular sort of audience;

they are thus much closer to embodying rationales than they are to recounting courses

of reasoning. As a consequence, even if such a judgment turns out to be good according

to whatever standard there is for goodness in legal judgments, that does not mean that

the judge reasoned well in rendering it. For instance, the judge may have labored far

longer to reach a decision than another might have.

Dancy has also looked to literature for models of good reasoning. He highlights

a passage from Anthony Trollope’s The Eustace Diamonds:

One of the elder Fawn girls had assured her that under no circumstances

could a lady be justified in telling a gentleman that he had spoken an

untruth, and she was not quite sure that the law so laid down was right.

And then she could not but remember that the gentleman in question

was Lord Fawn, and that she was Lady Fawn’s governess. But Mr.

Greystock was her affianced lover, and her first duty was to him. And

then, granting that she herself had been wrong in accusing Lord Fawn

of untruth, she could not refrain from asking herself whether he had not

been much more wrong in saying in her hearing that Mr. Greystock was

not a gentleman. And his offence had preceded her offence, and had

caused it! She hardly knew whether she did or did not owe an apology

to Lord Fawn, but she was quite sure that Lord Fawn owed an apology

to her. (quoted in Dancy 2018, 57)

I have no objection to this passage as a plausible example of active reasoning. But

Dancy goes on to comment: “The fact is that in the present deplorable state of the

theory of reasoning we are entirely unable to reveal the rational coherence of this train

of thought.” Is rational coherence what trains of thought should be expected to have?

The series of thoughts that occur to a reasoner need not be a matter of coherence or

implication, and whether this train of thought is a good one depends on various

material facts of her situation—for instance, whether there is some other important

detail that she has forgotten or overlooked, how immediately these thoughts occurred

to her, and whether they are true. It is doubtful that there is anything very determinate

for a philosophical theory of reasoning to say about what sort of things it would be a

mark of a good reasoner to remember or notice, other than that a good reasoner will

tend to remember and notice things which need to be remembered and noticed, for the

adequate addressing of her problem.

It seems to me, then, that the distinction between reasoning and rationale should not be

ignored, if we would like to do justice to the assessment of reasoning as good or bad.

We need to appreciate that reasoning and rationale are subject to distinct standards—to

standards of meansend efficiency and of argument, respectively. They are, of course,

related, since what one seeks, in reasoning, is to respond reasonably (on the basis of

some rationale) to one’s question, but this does not mean that reasoning itself is

argument-evaluable.

1.6 Reasoning Well

What does it mean to say that reasoning is means-end evaluable? As I pointed out in

Section 1.3, this point needs to be understood carefully. For reasoning, as a kind of

inquiry, proceeds from doubt, and the path to an answer to one’s question cannot in

general be known to the reasoner in advance. So reasoners are not necessarily to be

faulted for considering dead-end possibilities, for instance, even though the

consideration of dead-end possibilities is, in a sense, not conducive to answering one’s

question. In what sense, then, can a reasoner be reasoning well in considering a dead-

end possibility? Intuitively, a reasoner will be reasoning well if what she is considering

is a sensible thing to consider. But what makes something a sensible thing to consider?

In thinking about the assessment of reasoning, we must keep several points in view,

some of which we have encountered above. First, someone reasoning badly might

happen across not only a correct answer, but a correct rationale, by luck or by accident.

Thus:

(7) Successful reasoning is not necessarily good.

Conversely, someone reasoning well might not attain an answer, whether by some sort

of accident or some factor such as lack of time or difficulty of question. Thus:

(8) Good reasoning is not necessarily successful.

To these points, we can add one that Aristotle makes in giving his own account

of excellence of deliberation (εὐβουλία):

[O]ne person can achieve it by deliberating for a long time, while

another manages it quickly. The former case, then, still won’t count as

deliberative excellence; rather, deliberative excellence is correctness as

to what one should achieve, and the way in which, and when, all in

accordance with what is beneficial. (1142b27-29)

In other words, there is some kind of requirement of efficiency on good reasoning. I

will frame this requirement fairly weakly:

(9) It may redound negatively on the quality of reasoning if it takes longer than

it might have.

The grounds for this observation are, it seems, multiple. In the case of deliberation, or

practical reasoning, it has an instrumental basis. One who takes too long to deliberate

about some prospective good may thereby fail to attain it at all, since the opportunity

for it may be gone by the

time he has figured out how to attain it, and anyway he may take so long deliberating

that he loses out on other opportunities.

But some sort of requirement on efficiency plausibly holds in other domains of

reasoning as well. All else being equal, one’s reasoning is shown to be better if one

answers one’s question more quickly than one might have (or than other people do), at

least if this is not merely attributable to accident or luck. I do not mean to imply that

this requirement is very stringent. Moreover, its ground may, in some cases at least, be

indirect. The issue with a prolonged episode of reasoning may be, for example, that it

was delayed by difficulty, or even bias, and it reflects badly on one’s reasoning that one

struggled, e.g., to think of relevant possibilities. Think back to Dancy’s example of the

judge. A seasoned judge who is not able to recall any relevant precedents is to that

extent reasoning badly, but the same may not be true of a judge who takes longer to

reason because he is at liberty to do so in a more leisurely manner; he has, say, a light

day.

So what faster reasoning at least sometimes seems to show is that the person reasoning

is more capable as a reasoner. This suggests that in our treatment of what it is for

reasoning to go well, we might look in the direction of the notions of capacity or

disposition. (I will say more below about the distinction between these two notions,

and why I appeal to disposition in particular in my own treatment of the assessment of

reasoning.)

Aristotle has recourse to such notions in his own treatment of what is required for

someone’s action to be deemed skillful:

One can do something literate both by chance and at someone else’s

prompting. One will only count as literate, then, if one both does

something literate and does it in the way a literate person does it; and

this is a matter of doing it in accordance with one’s own expert

knowledge of letters. (2002, 1105a23-27)

The problem Aristotle is confronting is parallel to our own. Production (ποίησις) aims

at bringing about a certain result—for instance, a house. A house may be better or

worse relative to the standards there are for goodness in houses: whether it provides

shelter, is stable, and is comfortable. But what is the relationship between the goodness

of the product and the goodness of the production? The latter cannot simply be a

function of the former, for reasons parallel to (7), (8), and (9). We see Aristotle here

appeal to the producer’s knowledge, a kind of capacity to produce. The move aptly

captures the fact that—although a skilled producer will occasionally make a bad

product and an unskilled one a good product—it is no accident for a skilled producer to

make a good product.

Yet we should not simply insist that good reasoning is reasoning by someone disposed

to reason well. This would tie good reasoning too closely to the disposition to reason

well. It is intelligible that someone might say, for instance, that she was reasoning

unusually or uncharacteristically well during her mathematics exam on a given day.

Thus:

(10) Someone without a disposition to reason well might nevertheless reason

well. The possibility envisioned here is distinct from (7), which concerns the bearing of

accident or luck on the quality of reasoning. Someone who succeeds merely by

accident or luck seems not thereby to have reasoned well; what is now being

envisioned is that someone who is not disposed to reason well may still reason well.

The converse point holds too:

(11) Someone with a disposition to reason well might nevertheless fail to

reason well. That is to say, a good reasoner still might slip up on a particular occasion.

(This possibility is in turn distinct from (8).) So having a disposition to reason well is

neither necessary nor sufficient for reasoning well, and good reasoning cannot simply

be identified as that which proceeds from a disposition to reason well.

Still, I want to insist, the relationship between good reasoning and the disposition to

reason well is not merely contingent or extrinsic. It is important that the cases marked

out by (10) and (11) are uncharacteristic of the persons in question. When someone

who is not generally a good reasoner reasons well, this is, it seems, because she is

reasoning like someone who does generally reason well. She is, for example, noticing

what a good reasoner would notice, even if it is not the sort of thing she would

typically notice; or her recall of relevant facts is comparably swift. Thus, I want to

propose the following account of good reasoning:

(12) A person reasons well concerning a question Q when she reasons in a

way characteristic of someone who is disposed to answer questions like Q correctly

and efficiently.

Several comments on this account are in order.

(a) “disposed”: A disposition to φ is a tendency to φ, perhaps in particular

circumstances. For instance, the brittleness of glass means that it is disposed to break

when struck. This point requires some care. The dispositions for certain kinds of

reasoning, for instance technical reasoning, only need be exercised when one wishes to

exercise them, e.g., when one wishes to construct a house. It might naturally therefore

be called a capacity rather than a disposition. Similarly, it is usually at one’s discretion

whether one takes up a given mathematical question. But I take practical reasoning—

and perhaps, though I will not say more about them, certain fundamental kinds of

theoretical reasoning—to differ in this respect. I say more about this in the next

section.

(b) “to answer”: A subject A answers some question Q if A φ-s on the basis

of a rationale for φ-ing where φ-ing is a response to Q. For example, one might

respond to the question whether

there are infinitely many prime numbers by believing that there are. One might respond

to the question of what to do by dancing a tango or by pressing the detonator.

In marking out these last possibilities, of answering a practical question by

acting, I have assumed the controverted Aristotelian Thesis according to which the

conclusion of practical reasoning is sometimes an action, rather than merely an

evaluative belief or an intention. This assumption will not substantially affect the

arguments I make concerning practical rationality in this dissertation; I assume the

Aristotelian Thesis only because I must make some assumption about what it is to

respond to a practical question, and the Aristotelian Thesis is, syntactically at least, a

simple one, not to mention one I think correct.

and efficiently as regards, say, mathematical questions but not as regards literary ones.

Good physicists need not be good biologists. Good logicians might lack good sense in

practical matters. Accordingly, I will assume that the dispositions in question are

distinguished roughly by classes of questions to which they are responsive, and I will

refer to dispositions to answer questions correctly and efficiently as species of

rationality.

(d) “correctly”: I say that A answers Q correctly if the rationale on the basis

of which A φs is a good rationale. What this means is clear enough in the case of

mathematical or technical rationales. (Is it a proof of the theorem according to the

standards of mathematical logic? Will a house built according to these plans stand?) It

will be less clear, or anyway more controversial, in the more general theoretical and

practical contexts. I postpone the treatment of goodness in or

soundness of practical rationales to Chapter 4, and do not mean to treat goodness in

other rationales

at all.

Correctness is, at least in some respects and in some cases, gradable. In some

domains at least, there may be answers that are better than others, as well as rationales

that provide better support for a given answer than other ones do. On such bases, the

quality of reasoning will likewise be gradable.

(e) “efficiently”: A answers Q efficiently if A responds on the basis of a

rationale quickly and without wasteful effort, given the nature of her question and her

situation. As noted above, this requirement introduces a comparative dimension to the

assessment of reasoning. Aristotle approves of the adage that one should deliberate

slowly, but he also notes that one’s deliberation might be less than excellent in virtue of

its taking too long. Plainly there is no contradiction here. How quickly one reasonably

can, and therefore should, reason depends on the question and the situation. Some

questions are more difficult than others, and it is necessary and reasonable to devote

more time to answering them. In some cases this might mean that, compatibly with the

reasoning’s being good, years may be devoted to answering a question. Sometimes, for

instance in practical matters, a rough answer is needed immediately, and one must act

even without deliberation. Much hangs, then, on the proviso: ‘given the nature of her

question and her

situation’.

Efficiency is, like correctness, gradable. One can reason more or less efficiently,

and one’s reasoning will accordingly be better or worse.

There are, as I have indicated, different dimensions of efficiency—for instance,

speed and resource consumption. Pragmatic considerations in a particular case (the

urgency of a question or the availability of resources, for example) may dictate that one

or the other of these dimensions is especially pertinent, but this need not always be so. I

do not assume that there is a single measure by which all instances of reasoning can be

neutrally compared. It may be that the most one could say is that one instance of

reasoning is better in some respects but worse in others than another instance of

reasoning.

(f) “in a way characteristic of someone who …”: There may be various

ways in which good reasoners reason, especially but not only because good answers

and good rationales need not be unique. Accordingly, to reason well, one only need

reason in a way characteristic of someone who is disposed to answer the relevant kinds

of questions correctly and efficiently.

I take it that there is no general theory, from without the perspective of a good

reasoner, from which the best ways of reasoning can be derived—for instance, a theory

from which it follows that it would be sensible to try proof by mathematical induction

in a given case. When we identify someone’s reasoning as going relatively well or

badly, we draw on our own conception of what sort of things it would be helpful to

remember and notice in a particular case.

Now, there are familiar troubles with appealing to an ideal subject in giving accounts

of,

e.g., what reasons an agent has (cf. Smith 1995). Namely, there are cases in which a

non-ideal agent should not act as an ideal one would, precisely because the non-ideal

agent knows herself to be non-ideal. This sort of possibility leads Michael Smith to

suggest that what an agent has reason to do is, rather, what that agent’s ideal self would

advise her to do—a proposal that faces its own problems. I do not believe that parallel

worries afflict the present account, however. It may be that a non-ideal reasoner should

not reason as an ideal one would (perhaps she should consider possibilities that an ideal

reasoner would know immediately to overlook), but the present account is not about

how anyone should reason; it is about what makes reasoning go better rather than

worse. The non-ideal reasoner who reasons as she, given her limitations, should will

still be, in turn, reasoning in a way that is worse than the way the ideal reasoner

reasons. That seems to me to be the right result.

(f) “disposed to answer questions …”: I specify the disposition in question as a

disposition to answer questions, rather than as a disposition to reason well. Such a

disposition is, but is not only, a disposition to reason well, because a disposition is

general in character, and it is responding to some but not all questions that requires

reasoning. Such is the lot of finite reasoners. That thought is consistent with the

following:

(13) It is a limiting case of the exercise of rationality to respond without

reasoning to some questions.

This is particularly important, I take it, in the case of practical rationality, where one’s

situation may demand a virtually immediate response. It is characteristic of the

practically rational person to reason well when one’s situation is complex and calls for

judicious reasoning, but also to ‘see’ what to do when that is necessary and possible.

We must distinguish responding without reasoning from responding without a

rationale. As Aristotle says, the virtue of courage sometimes drives out deliberation

(2002, 1117a18-23), but that does not mean that it occurs without what he calls

προαίρεσις (choice, decision), which is based on a λόγος, a rationale. A disposition to

respond without any rationale is what Aristotle calls εὐστοχία—“a mere capacity to

guess correctly” (literally, ‘shooting well’). In human beings, such

a disposition would inevitably be limited, for humans simply cannot guess correctly

what to think and what to do in every case. Sometimes they must reason, if they are

reliably to hit the truth, which is regularly at some distance from what they already

know.

1.7 Conducting Oneself

The general idea I have developed in the previous section is that we must appeal to the

notion of a disposition in an account of what it is to reason well, due to the ad hoc way

in which reasoning must proceed—by way of noticing and remembering what is

salient. In this section, I want to turn to the topic of practical rationality. In particular, I

want to flag a further way in which practical rationality specifically involves a

disposition to remember and notice salient features of one’s situation. In the next

chapter, I will introduce an account of practical rationality that coheres with this datum.

I will introduce the core idea by way of an example. Suppose that among the

things I reasonably care about is that my guests enjoy good hospitality at my hands,

and I am currently hosting a party. One of my guests is, for some reason or other,

uncomfortable, and there is something that could be done about that. But although that

fact would stand out to a more percipient and provident host, I do not notice his

discomfort, perhaps because I am thinking about some other aspect of my party—

whether it would be a good time to put the little quiches in the oven, say—while also

trying to follow the conversation. I might be coming up with as fine an answer to that

question as you like; based on how I am deliberating now, I am liable to put the

little quiches in the oven at precisely the ideal time, as far as the little quiches are

concerned. Yet I do not notice, and therefore do not address, my friend’s discomfort.

The intuition I want here to register is that in this example there is something wrong

with my practical thinking. There is something wrong, even by my own lights; if I were

apprised of the fact that my friend is uncomfortable, that is, I would concede that I was

reasoning badly. The answer I propose is that while I am doing a fine job of answering

the question when to put the little quiches in the oven, I am doing a poor job answering

the question what to do. This latter question is, for human beings, unavoidable. In this

way, practical rationality is not simply a capacity to answer practical question, but a

disposition to answer what we can call the agent’s question.

In what sense is the agent’s question unavoidable? Anselm Müller (2004a)

distinguishes acting, as it figures in the notion of acting well, from intentional action.

He interestingly suggests that either acting well or acting badly is predicable of any

“sane adult human being”: You may produce a video that shows a person repairing a

car as opposed to repairing a motor bike, or to climbing a staircase. But what could

acting be ‘opposed to’? How would you go about producing a video that showed a

person acting? If any video can do the job, then any video can—as long as it features

some sane adult human being. That is, the neutral component of the concept of acting

reduces to the voluntariness that is characteristic of the way human beings operate.

There is nothing else that different ‘cases of acting’ have in common. (Müller 2004, 18)

The notion of voluntariness in play here derives from a certain tradition of theorizing

about ethical responsibility. It is a kind of axiom for Thomas Aquinas that only what is

voluntary can be a sin, or—to put it in a less theological register—a fault. This is not

meant to be a controversial thesis restricting our understanding of what people can be

responsible for to the boundaries of a preconceived, say physiological, notion of the

voluntary. It is, rather, a sort of implicit definition of the voluntary. Anscombe here

explains Aquinas’s position:

[Aquinas] asks whether voluntariness can be there without any act [of

the will] at all. He says that what is from the will is what is called

voluntary. But this ‘from something’ has two applications. In one way

the ‘from’ is direct because something proceeds from something else

which is active, as heating proceeds from what is hot. In the other way

it is indirect, from the very fact of not acting, as one says that the

sinking of a ship comes from the pilot inasmuch as he was not engaged

in piloting it. But it has to be realised that what follows from lack of

action does not always go back to an agent as cause from the mere fact

of his not acting, but only when he can and ought to act. (2008, 130-

131)

The voluntary, accordingly, extends not just beyond the intentional but even beyond the

known. The ship’s crashing is imputed to the pilot, rather than to other members of the

crew, because the pilot could and should have been attending to its path, even though—

since he was not—he did not know that the ship was about to hit an obstacle. The pilot,

doing whatever else he was doing, was, as I will put it, conducting himself badly.

Similarly, in the example I have introduced, I am conducting myself badly because I

could and should notice my guest’s discomfort, but I do not. Any sane adult human

being is conducting himself or herself either well or badly—even one who is asleep.

(Perhaps it is her job not to be asleep right now.) The assessment of conduct must make

reference to this wider context, and not simply to, say, the particular grounds on which

a person does act, in virtue of which her action is the sort of intentional action it is. For

the agent may not be acting intentionally at all, or may be doing something which

appears to be all right and well-motivated in itself but amounts to negligence of some

other duty in the present context.

I want to frame this fact, about the universal imputability of either the description

‘conducting oneself well’ or the description ‘conducting oneself badly’, in terms of the

unavoidability of the agent’s question—since what the agent can and ought to do is

what determines how she is conducting herself. For mature human beings, the agent’s

question is effectively always in the offing. Every human being is, by her conduct,

offering a tacit answer to this fundamental question, an answer that will, or will not,

measure up to whatever standards there are for practical rationales. Practical rationality

then must be not just a disposition to answer specific practical questions—such as

when to put the little quiches in the oven—but it must be a disposition to notice which

specific practical questions are pertinent to the answering of the unavoidable agent’s

question.

In the next chapter, I will develop and defend an account of practical rationality

that gives it an appropriate place in good human conduct.

1.8 Conclusion

The perspective on the nature of rationality which I have introduced in this chapter is

distinct from, though at least in some measure compatible with, existing perspectives.

Rationality is often thought to consist in one’s satisfying requirements on one’s

attitudes, whether at a given time or over time (Broome 2013, 149-191). Some

philosophers take it to consist in one’s responding correctly to reasons (Kolodny 2005;

2008). I do not want to suggest that there is nothing to these

ideas, though I do want to insist that there is an important dimension which such

accounts leave out. Broome writes, for instance:

Suppose your attitudes satisfy all the requirements that apply to them.

You would not be fully rational if they did so purely by coincidence. To

be fully rational, you must have dispositions that would lead you to

satisfy the requirements in other circumstances too. Rationality requires

you to have those dispositions. But I shall not try to formulate

requirements on dispositions. (2013, 149).

We can say something even stronger, perhaps. The world might conspire against a

rational person in such a way that he is unable to answer the questions to which he

addresses himself, or answers them wrongly. Nevertheless he may have whatever

species of rationality is in question. It is simply unfortunate that circumstances are such

that that disposition’s normal exercise is impossible. Conversely, as Broome indicates,

one might satisfy requirements of rationality (or manage to respond correctly to

reasons) despite not being rational—having no disposition to satisfy those requirements

(or respond correctly to reasons). One might even have such a disposition and not be

rational, because one is not disposed to satisfy the requirements of rationality (or

respond correctly to reasons) efficiently.

Rationality as it is presented here is an excellence not of the states one happens to be

in at one time or another but of one’s overall disposition for mental conduct. A species

of rationality is a disposition to get things right within a certain domain. My focus in

the remainder of the dissertation will be on practical rationality, or wisdom: a

disposition to get things right in the domain of things to be done.

Chapter 2: The Virtuous Agent’s Reasons

[The thesis that the V person chooses V acts qua V acts] is very roughly

right, but only if “qua V” is read de re. We say that the agent did the

generous (e.g.) thing because it was the generous thing to do, and we

understand what this means because we understand what it is about the

situation and the action that makes this action in this situation

something that would seem to a generous person the appropriate thing

to do. It will follow from this that the philosophical understanding of

the various virtues will require some, at least, of the understanding that

comes from having the virtues: which is of course what Aristotle holds.

(Williams 2006, 193)

2.1 Introduction

In her little book Natural Goodness (2001, henceforth NG), Philippa Foot locates the

goodness of virtue in its place in human life. In particular, Foot argues that the

goodness of living things (their goodness as the kind of thing they are, or their natural

goodness) is measured by reference to their life form (the living and being

characteristic of their species). So too it is, Foot suggests, with human beings. But the

vital capacities of human beings include not just digestion, locomotion, and sight, but

also reason, including practical reason. Thus, the characteristics of the excellent

disposition of practical reason—that is, practical rationality—are determined by the

role of practical reason in human life, just as the characteristics of excellent canine

sight are determined by the role of sight in canine life. On such a conception, Foot

thinks, practical rationality will be seen to consist in the possession and exercise of the

virtues, which she takes to be habits of responding to certain classes of reasons for

acting, in particular to those classes of reasons for acting to which humans need to

respond, if they are to attain their characteristic good. To be a practically rational

human being is to be a good human being, and to act rationally is to act well.

Thus NG accords a central place in the ethics of virtue to human nature. It

amounts to a defense of neo-Aristotelian naturalism, for Foot encapsulated in the thesis

that:

(1) Moral judgments are judgments of the natural goodness and defect of

human practical reason. (Cf. NG, 4)

This is supposed to be a claim about the logical character of what would familiarly be

recognized as moral judgments—judgments concerning good and bad, right and

wrong, and what ought to be done. It is meant to apply not just to the evaluation of

particular actions, but also of emotional reactions, persons, and ways of life. So

examples would include “You should go, since you promised,” “She’s a good person,”

and “What he did was wrong.” In Foot’s usage, moral judgments are distinguished

from prudential judgments—but she ultimately argues that they are on a par, and she

holds that prudential judgments too are judgments of the natural goodness and defect of

practical reason (cf. NG, 11).

In this chapter, I brush up against one prominent objection to the role of human nature

in Foot’s account. Namely, as many philosophers have read her, Foot takes the virtuous

person’s knowledge of human nature to provide her with her reasons for acting well.

Call this view derivationism:

(2) The virtuous person’s ultimate reasons for acting well are given in

judgments of natural goodness of human beings (or in judgments about the

human life form, to which such judgments of natural goodness are

internally related).

Derivationism has seemed to philosophers to be problematic in various ways. For

instance, arguably, there are few agents who possess such knowledge, which also

seems to provide the wrong sort of reason to perform such virtuous actions as keeping

a promise or helping a friend.

Is Foot a derivationist? I will suggest that Foot is at most ambivalent about

derivationism. Certain features of NG might seem to require derivationism, and at

points Foot seems to endorse it, so it is no surprise that commentators have taken her to

do so. However, I want to argue, neither Foot’s overall argument, nor her response to

skepticism, nor her conception of virtue, requires derivationism. The source of the

confusion, I suggest, is a failure on the part of Foot and her readers to distinguish

between two candidate sorts of reasons. Once this distinction is made clearly, Foot’s

conception of virtue as a disposition to respond to certain classes of reasons presents an

alternative to derivationism which has much to recommend it even outside its

importance in giving a more defensible reading of Foot.

2.2 The Basic Argument

NG’s pivotal moves draw from two sources: Michael Thompson’s conception of life

and Warren Quinn’s argument against end-neutral conceptions of practical rationality

(Thompson 1995; Quinn 1995).

Foot builds upon Michael Thompson’s account of Aristotelian categoricals, or

what he calls natural-historical judgments. Aristotelian categoricals are general

propositions of such forms as

“Ss are F”, “The S is F”, or “An S is F”—for instance, “Olive sea snakes have a

venomous bite”. Thompson argues that this form is distinctive in being irreducible to

familiar forms of logical generality. Although natural-historical judgments have some

sort of general character, they do not assert that all, or even most, Ss are F. They are

rather about what is typical for or characteristic of Ss. As Foot puts it when she

introduces Thompson’s notion, natural-historical judgments say of

Ss how they live (NG, 28). Such statements articulate what Thompson calls the “life

form” of Ss. One who knows various Aristotelian categoricals about Ss thereby knows

something of the “natural history” of Ss, the kind of thing narrated in nature

documentaries (Thompson 1995, 280291). A central theme of Thompson’s treatment is

that knowledge of the life form of Ss, of the sort expressed in natural-historical

judgments, is presupposed even in thoughts about individual Ss, since I can only

recognize a part of an organism as (say) an arm (rather than a leg, a tail, a deformity, or

a symbiote) if I know that this is the sort of organism that has arms (Thompson 2004,

51-54).

Thompson’s account of natural-historical judgments points toward a definition

of natural goodness and defect. If Ss are F—if it is part of the nature of Ss to be F—

then it is naturally good for individual Ss to be F, and in an S that is not F, this is a

natural defect. Hence a dog, but not a human, without four legs, is defective

(Thompson 1995, 295-296; 2004, 54-56). Part of the interest of such an account is that

it promises to go some way toward short-circuiting familiar skepticism about the

reality, knowability, and normativity of nature. In establishing a connection between

what sort of creature something is and what it is for it to act well or badly, it establishes

a connection between what is the case and what ought to be the case, between fact and

value. And the homey examples Thompson and Foot adduce are tailored to suggest that

knowledge of a thing’s nature is hardly rarified knowledge of a je ne sais quoi, but is

rather eminently familiar and

attainable. The account reminds us of the utter facility we have, in speaking both of

human and of non-human life, with the natural-historical grammatical form, and it

makes immediately and readily available to us objective talk of goodness and defect—

goodness and defect independent of the values and interests human beings happen to

have.

It is within this framework of natural goodness that Foot defends neo-

Aristotelian naturalism. For—she plausibly (if not uncontroversially) claims—practical

reason, the capacity to act for reasons, is just another power of human beings. This

power can be, as it were, disposed in one way or another. In what I take to be a key

passage (and to which I will later refer as the key passage), Foot suggests that to

possess a virtue just is for one’s capacity of practical reason to be disposed to respond

to certain sorts of reasons:

What … distinguishes a just person from one who is unjust? The fact

that he keeps his contracts? That cannot be right, because circumstances

may make it impossible for him to do so. Nor is it that he saves life

rather than kills innocent people, for by blameless mishap he may kill

rather than save. ‘Of course,’ someone will say at that point, ‘it is the

just person’s intention, not what he actually brings about, that counts.’

But why not say, then, that it is the distinguishing characteristic of the

just that for them certain considerations count as reasons for action,

and as reasons of a given weight? Will it not be the same with other

virtues, as for instance for the virtues of charity, courage, and

temperance? Those who possess these virtues possess them in so far as

they recognize certain considerations (such as the fact of a promise, or

of a neighbour’s need) as powerful, and in many circumstances

compelling, reasons for action. They recognize the reasons, and act on them. (NG,

12, emphasis in original)

In this way Foot applies Warren Quinn’s proposal. Quinn (1995) asked why, if practical

rationality could be put to bad ends, it should be thought so important. I find that

argument questionable. Whatever one thinks of it, though, I take Quinn’s proposal to

be attractive simply on the assumption that practical reason is a vital power. If practical

reason is indeed another power of human beings, then one need not suppose that the

standards for its good exercise are given a priori, in advance; it is a mistake “to think

that moral action has to be brought under a pre-established concept of practical

rationality” (NG, 10). Instead, one can hold that practical rationality, the good state of

practical reason, its virtue, is itself sensitive to certain ends and, indeed, that “there is

no criterion for practical rationality that is not derived from that of goodness of the

will” (NG, 11).

Foot says that the conceptual structure of natural goodness “remains intact”

even as one turns to the richer and more complicated human good. Human beings are

good when they are such as to attain their characteristic good well. This will mean

possessing such virtues as there are, since virtues just are good states of practical

reason. Foot thinks that the human virtues turn out to look rather like the traditional

catalogue:

[I]f we ask whether Geach was right to say that human beings need

virtues as bees need stings …, the answer is surely that he was. Men

and women need to be industrious and tenacious of purpose not only so

as to be able to house, clothe, and feed themselves, but also to pursue

human ends having to do with love and friendship. They need the

ability to form family ties, friendships, and special relations with

neighbours. They also need codes of conduct. And how could they

have all these things without virtues such as loyalty, fairness, kindness, and in

certain circumstances obedience? (NG, 44-45)

Those who act well, from the virtues, are naturally good in respect of their capacity of

practical reason, while those who act badly, out of accordance with the virtues, are bad

in respect of that capacity. They are practically irrational, and this is a kind of natural

defect. This is the thesis of neo-Aristotelian naturalism.

I take what I have just sketched to be, in outline, Foot’s argument that acting

well requires acting in accordance with the virtues and that it is practically rational to

act well. Notice, however, that derivationism plays no role in it. Virtue is conceived as

the disposition to act for certain kinds of reason; the reasons which characterize virtue

are those on which human beings need to be disposed to act, if they are to achieve their

characteristic good. The reasons on which they act are, for instance, “I promised her I’d

be there” and “He could use my help”—not “Human beings keep promises” or “It is

good for human beings to keep promises” or “Human beings ought to keep promises”

or “Human beings need to keep promises because otherwise they could not achieve a

great deal of human good” (cp. Anscombe 1969, 73-75; NG, 15). All of these latter

claims are, on my non-derivationist reading of Foot, true, but they are not among the

virtuous agent’s reasons for keeping a promise.

2.3 The Derivationist Alternative

Nevertheless, the reading of Foot as a derivationist predates even the publication of

NG, beginning with John McDowell’s celebrated “Two Sorts of Naturalism.”

McDowell invites us to imagine that a wolf acquires the capacity for practical reason

(2001b, 169-173). Since reason is (McDowell says) a capacity for reflection and

criticism, this acquisition would enable the wolf to call into question the normative

authority of such truths of wolf nature as that wolves hunt cooperatively and share their

spoils with the rest of the pack. So McDowell evidently thinks that, for Foot, truths of

such form have an essential role in the agent’s deliberation, such that their normative

authority would be undermined by their being questioned. That is, she is a

derivationist.

Now, McDowell’s paper is—appropriately, since it was published before NG—

ambivalent about ascribing the view it criticizes to Foot herself. It is plausible to

suppose that its target was simply one possible way of developing and applying to

ethics the account one finds in Michael Thompson’s “The Representation of Life.”

Whatever the case may be, readers of Foot after NG have tended to see McDowell’s

wolf as posing an urgent challenge to Foot’s account, and they have often read her as a

derivationist. To take some especially explicit examples:

The virtuous person [for Foot] is one who makes true judgments about

what is choiceworthy for human beings…, and these judgments are

grounded in her recognition of the true facts about human needs and

goods. (Frey 2018, 60) [F]or Foot, theoretical knowledge of human

nature is supposed to play the role of a major premise in practical

reasoning. (Ometto 2023, 136)

Not all readings of Foot ascribe derivationism to her so baldly. Thompson

himself has developed on Foot’s behalf a response to McDowell’s example of the

rational wolf (2004; 2022). He agrees that it would be problematic to give a role to

empirical knowledge of human nature in an account of the reasons the virtuous agent

acts on, but he argues that there remains space for holding that the virtuous person acts

from non-empirical, a priori knowledge of the human life form. Lest one doubt that a

priori knowledge of the human life form is possible, Thompson invites us to consider

the activity of thinking. Thinking is (Thompson claims) a vital activity, and it is also

an activity in which the subject self-consciously represents herself as thinking. This

self-conscious representation draws on a conception of a life form in just the way that

all representation of vital activity draws on a conception of a life form. So, simply in

thinking, one must have some knowledge of the human form. In this sense, as

Thompson puts it, self-consciousness is a kind of form-consciousness (2004, 66-68).

His proposal is that there may be also, and similarly, a kind of a priori practical

knowledge of the human life form (2004, 72; 2022, 74-79). The idea then would be

that, since practical knowledge is “the cause of what it understands,” the rational wolf’s

skeptical question cannot get any purchase, cannot find any gap.

This is, no doubt, an interesting proposal, and it has been an influential one—it

has been taken up, in different ways, by readers of Foot such as Micah Lott (2012b;

2014), John HackerWright (2013; 2021), and Matthias Haase (2018). As I read them,

authors vary in how explicitly they avoid derivationism (cp. Thompson 2022, 76-77;

Hacker-Wright 2013, 85). Often, knowledge of human form is said to be somehow

“implicit” in the virtuous agent’s perspective. There is, I think, something right about

that, to which I will come later, but my immediate interest is that this elaborate reply to

McDowell does not begin by questioning his very fundamental assumption that Foot is

a derivationist. As will emerge below, I take questioning this assumption to be essential

to understanding how Foot herself means to answer the skeptic, whose question is the

same as the rational wolf’s.

Other Foot-adjacent philosophers are less invested in the Thompsonian framing

but likewise think that Foot goes awry in holding that knowledge of human nature

could be among one’s reasons for acting well. Rosalind Hursthouse was one of the

earliest authors to develop an account of virtue indebted to Foot but responsive to

McDowell’s objection (Hursthouse 2001, 165166). Other recent examples include Frey

(2018) and Müller (2018), who draw from Thomas Aquinas’s account of synderesis for

the idea that an immediate and distinctively practical recognition of human goods is the

source of reasons for action.

My immediate focus is not the merits of these various alternatives but rather the

reading of Foot to which they are each responsive. Since McDowell’s “Two Sorts of

Naturalism,” derivationism has been the standard reading of Foot’s late work. This is

not, I think, an accident. Notwithstanding the non-derivationist sketch of NG I gave in

Section 2.2, certain aims of and considerations within NG seem to support

derivationism; I consider these in Section 2.5. Before that, in the next section, I

consider some points where Foot simply seems to affirm derivationism explicitly. My

suggestion will be that we need to understand these moments in terms of a distinction

which is present in the text yet which Foot and her readers fail to draw explicitly.

2.4 Situational Reasons and Evaluations

Foot writes:

[A] human being can and should understand that, and why, there is

reason for, say, keeping a promise or behaving fairly. This last may

seem a tall order, but this human understanding is not anything hard to

come by. We all know enough to say, ‘How could we get on without

justice?’, ‘Where would we be if no one helped anyone else?’, or ‘How

could we manage if there were no way of making decisions for us all?’

(NG, 12)

These rhetorical questions, which Foot envisions competent agents as raising, are

meant to call attention to anthropological considerations similar to those Anscombe

(1969) insists upon in her account of promising. Human beings need to be just toward

one another, to cooperate with each

other, and to coordinate with one another, or else they could not attain many of their

goals. A great deal of human good depends on virtuous conduct, or some

approximation to it. (For further discussion of Anscombe’s account of promising, see

Section 3.2.)

Now, one could of course hold that “a human being can and should understand

that, and why, there is a reason”—say, the fact that p—for some virtuous conduct,

without holding that p should be her reason for performing that virtuous conduct. The

place of and need for virtuous conduct in human life might just be the sort of thing of

which one should be aware, even if it is not what motivates one to virtuous conduct.

This is what Hlobil and Nieswandt, in defending a reading of Foot similar to mine, say

about this passage (2019, 1511-1512).

Not all of Foot’s remarks are susceptible of this kind of reading. For instance,

she writes: “One who is the subject of a true moral judgement does not always do what

it says he should do, since he may not recognize its truth, and may not act on it even if

he does” (NG, 18, emphasis added). Here, the moral judgment, which is a judgment of

natural goodness, is said to be the sort of thing which agents act on (or are criticizable

for failing to act on). If moral judgments are judgments of natural goodness and defect,

then Foot appears to affirm derivationism here.

I will consider how we should think about a remark like this in Section 2.6. For

now, I want to draw a distinction that will help clarify this material. When Foot refers

to the reasons on which a virtuous agent acts, she does not distinguish considerations of

two sorts, which I will identify by lists of examples. Situational reasons include such

considerations as that

(i) I promised to φ;

(ii) She needs X;

(iii) X is his;

(iv) φ-ing might damage my health.

Situational reasons are simply the kind at which Foot gestured in the key passage. They

are particular facts about the world, the disposition to recognize and respond to certain

classes of which characterizes one as possessing one or another virtue. She gave (i) as

an example of a reason associated with justice.

The class of what I will call evaluations consists of such considerations as that

(v) There is a reason for me to keep my promise;

(vi) It is good, when possible, to provide others with what they need;

(vii) I ought to respect other people’s property;

(viii) It is reasonable to avoid what may damage one’s health.

I take it that the label ‘evaluations’ for these judgments is natural; they each employ

evaluative or ‘normative’ terminology. Foot herself calls considerations such as these

evaluations of the human will. For Foot, evaluations all bear some immediate

inferential relation to natural-historical judgments, by reference to which the evaluation

of human beings as good or bad, non-defective or defective, is made. The ‘moral

judgments’ referred to in (1) are all evaluations.

I will employ this distinction between situational reasons and evaluations in arguing

that we need not read Foot as a derivationist.

2.5 For and Against a Derivationist Foot

Why has Foot been read as a derivationist? I will consider three aspects of NG which

seem to support to a derivationist reading. First, derivationism can seem to follow from

a requirement Foot places on any account of practical reason and morality: what she

calls Hume’s practicality requirement. Second, Foot aspires in NG to provide a

response to the moral skeptic, and it might be thought that she does so by affirming

derivationism. Finally, and I think most fundamentally, Foot’s inclusion of evaluations

among the virtuous agent’s reasons seems to commit her to derivationism. I will

consider these potential grounds for derivationism in turn.

(a) Foot presents her account as an alternative to the various forms of

subjectivism that had been entertained in the latter half of the twentieth century, and

while she believes that such forms of subjectivism were misguided, she also holds that

they aimed at addressing a genuine requirement on any account of morality and

rationality:

Their theories were devised to take account of something that really is a

feature of moral judgement: the ‘action-guiding’ character of morality,

which Hume had insisted on and taken as the foundation of his moral

philosophy. Morality, Hume has said, is necessarily practical, serving to

produce and prevent action, and I shall call this ‘Hume’s practicality

requirement’. Nor am I denying that his demand must be met. My

contention is rather that the theories I am attacking tried to meet it in the

wrong way. (NG, 9)

We have, then, the following requirement:

(3) Morality is practical; it is action-guiding; it serves to produce and prevent

action.

(3) appears to undergird derivationism because according to neo-Aristotelian

naturalism, moral judgments are judgments of natural goodness and defect. If moral

judgments are practical in the sense action-guiding, this seems to mean that they are

among one’s (or the virtuous agent’s) reasons for acting.

But it is important to appreciate that Foot seeks to make neo-Aristotelian naturalism

serviceable by way of a reorientation of our thinking about practical reason and

practical rationality. The lesson Foot says she draws from Warren Quinn is that it is a

mistake “to think that

moral action has to be brought under a pre-established concept of practical rationality”

(NG, 10). She accepts (3), but as she says, “Hume’s demand is met by the (most un-

Humean) thought that acting morally is part of practical rationality” (NG, 9). We can

see what she has in mind by this proposal in light of the sketch of the book’s argument

that I gave in Section 2.2. To be practically rational just is to be virtuous, and a virtuous

agent—so the key passage says—is one who recognizes and responds to the

appropriate class of situational reasons. Accordingly, it would not be arbitrary to call

those reasons moral reasons. That is how “acting morally is part of practical

rationality,” so it is also how, for Foot, “morality” is action-guiding.

To put the point differently, (3) is ambiguous. To say that “morality” is action-

guiding might mean that the situational reasons which characterize the virtues are, or

that moral evaluations are. We should not assume that Foot means to take (3) in the

latter manner, just because her subjectivist opponents do, for she says they tried to meet

Hume’s practicality requirement in the wrong way. To decide what Foot has in mind,

we need to attend to how her arguments actually function. When we do, we see that it

is not the case that morality is action-guiding because judgments of natural goodness

exert a kind of force on just anybody, but rather it is action-guiding because the

situational reasons which characterize the virtues are the reasons the virtuous agent is

disposed to act on.

(b) Foot takes up Gary Watson’s question of whether moral philosophy “can

establish an intelligible connection between [the appraisal of a gangster as a bad human

being] and what we have reason to do as individuals” (Watson 1997, 67, quoted in NG,

53). She wants, that is, to refute

the skeptic, who asks “why he should do that which the good person must do” (NG,

64). It might seem that derivationism is her way of answering this skeptic, just as it has

seemed to many that derivationism would be her way of responding to McDowell’s

rational wolf, whose question is,

after all, the skeptic’s.

Anselm Müller addresses the adequacy of Foot’s response to the skeptic

explicitly. He doubts whether Foot has succeeded in answering the skeptic who “denies

that he has been given any reason to do what he has been given reason to believe he

ought to do” (2018, 173). The response that should be made here parallels that which I

just made concerning Hume’s practicality requirement. Just as Foot does not mean to

meet Hume’s practicality requirement by showing that judgments of natural goodness

(concerning what human beings “ought to do”) exert a kind of normative force on just

anybody who believes them, Foot does not mean to answer the skeptic by showing that

he will be persuaded to act well once he has been given an account of human nature.

Hume’s practicality requirement is met in the “most un-Humean” way of arguing that

the practically rational agent just is a virtuous agent—an agent who acts on certain

kinds of situational reasons. That is how morality is “practical” and how in failing to

act well one is failing to respond to the reasons which there are. Moral reasons are

reasons for the skeptic too, and he is practically irrational for failing to act on them.

This means that the skeptic, if he is sincere (if he is really a gangster), is a bad person,

and argument may well not help him. Indeed, Foot writes: “In spite of recognizing the

force of Hume’s ‘practicality requirement’, we must allow for ignorance, for weakness

of will, and also for the phenomenon of shamelessness” (NG, 19). So a connection has

been established between the appraisal of a gangster as a bad human being and what

we have reason to do as individuals. But the connection is not one of the appraisal’s

being one’s (or the gangster’s) reason for not being a gangster. The connection is that,

if the appraisal is correct, then we have reasons for doing, de re, the things a good

human being does.

This reading makes sense of the character of Foot’s response to the skeptic.

Immediately before providing her response, Foot distinguishes between

(A)Reasons for acting, which we may call practical reasons.

(B) Reasons for believing, which we may call evidential or demonstrative

reasons. (NG, 63)

Then she says:

As philosophers, and therefore theoreticians, our job is of course to give

the second type of reason, arguing for or against the truth of a variety of

propositions that seem to involve special problems—like those, for

instance, about personal identity or the existence of an external world.

But among the many ‘philosophical’ subjects we find that of the nature

of practical reasons, and in this special case we shall have to give

reasons of type B for theses about reasons of type A. … When Gary

Watson issued the challenge described … above, he was asking a

question that belongs here

…. (NG, 64)

This accounts for the flat-footedness of Foot’s reply to the skeptic, which (depending

on how his challenge is to be understood) gestures backward to her treatment of the

nature of natural goodness or to her account of “the conceptual connection between

acting well and acting rationally” (NG, 64-65). In neither case does she take herself to

be providing a skeptic with what she calls practical reasons, appreciation of which

would get him to act well:

No doubt what our sceptic (especially if he be a gangster) really means to

insist is that we have not been able, in anything we have said, to touch his

desires; and if he is a dangerous person that may be what we shall most care

about. But the fact that we might hunt around for something that has a

chance of affecting his actions should not be taken as giving any support to

a philosophy that takes practical reasons to encompass only reasons of that

kind.

The practical reason why the gangster should keep his promise is the fact that he made

it, and if he were a good person, that would be enough for him. The consideration of

the place of promising in human life which Foot undertakes in her book is, by contrast,

an avowedly theoretical matter. It is not meant to be the sort of thing one can hand to a

non-virtuous agent to get him to act well.

saw in the previous section, that Foot often treats evaluations as reasons for action.

Such statements look to provide perhaps the fundamental ground for ascribing

derivationism to Foot. For they seem simply to say that moral judgments, which are

(according to neo-Aristotelian naturalism) judgments of natural goodness and defect,

are reasons for action.

But why does Foot treat evaluations as reasons for acting? With the distinction

between situational reasons and evaluations in hand, we will be required to

acknowledge that these are distinct claims:

(4) A virtue is (inter alia) a disposition to be motivated by a certain class of

situational reasons.

(5) A virtue is (inter alia) a disposition to be motivated by a certain class of

evaluations. In the key passage, Foot only claimed (4) to be true. Given (4),

we still need an argument for the truth of (5). In the next section, I will be

suggesting that (5) is not true, or at any rate needs qualification. Right now,

I will consider a reading on which Foot does provide a ground for inferring

(5).

Namely, Foot endorses what I call the qua-good requirement. This requirement

on an account of practical reason encapsulates Foot’s idea of what it is that

distinguishes human action from that of non-human animals. For non-human animals

too may be said, in a sense, to act for reasons. Foot approvingly cites Aquinas’s idea

that “while animals go for the good (thing) that they see, human beings go for what

they see as good: food, for example, being the good thing that animals see and go for

and that human beings are able to see as good” (NG, 56). Hence:

(6) Humans pursue goods qua good.

There are, I take it, a couple ways in which (6) might be taken to support (5).

One such way might begin from the question: what is it to see a good as good? A

natural answer would be that at least part of seeing a good as good is judging the good

to be good; one judges, of some action φ-ing, that φ-ing is good (or that one should φ,

or that there is reason for one to φ). This would be, for Foot, a judgment of natural

goodness. Accordingly, on this reading of (6), it seems that judgments of natural

goodness are among one’s reasons when one pursues the good qua good.

This is not exactly how Foot understands (6). But one might think that Foot’s

elucidation of (6) comes to more or less the same. In glossing the qua-good

requirement, Foot approximates

McDowell’s idea, mentioned above, that reason is essentially reflective or critical. It is

a power which enables a subject to “step back” from apparent reasons. Of any

candidate reason, it can sensibly be asked whether it is a reason and, if so, why:

Quite early on, a child learns that a ‘should’ needs a ground, unlike an order,

which may simply be reiterated or backed up by a threat.

… A human being a rational animal will ask ‘Why should I do that?’,

particularly if told that he should do something distasteful that seems to

be for the advantage of others rather than himself. And a philosopher,

even if in propria persona a respectable person inclined to keep his

promises and pay his debts, is honour bound to carry such questioning

to the limit. (NG, 56-57)

A mature human agent realizes, Foot says, that a ‘should’ needs a reason. And one

should act as morality requires. So the virtuous agent should be prepared to say why

she should do as morality requires. If this questioning is carried “to the limit,”

presumably it must terminate in a judgment of natural goodness and then in a natural-

historical judgment, a piece of knowledge about human nature; it might be asked why,

or how one knows that, such a thing is true, but if one grants Foot’s account of life,

then there is no problem in principle about answering such a question.

Even in this quotation, however, there is a contrast between how a “respectable

person” thinks about her reasons to act as virtue requires and what it is incumbent upon

a philosopher to consider. Foot, of course, would not suppose that every person is

bound to question the requirements of morality “to the limit.” Not everyone is required

to be a moral philosopher. We should recall that the point of the qua-good requirement

is to distinguish human from animal acting for reasons. And when we see how it does

so, we see that it does not require derivationism. I quote Foot at length on what it is for

a human being to see a good as good: [S]peech is crucial here in marking the difference

between animals and humans. We know what an animal is going after only by what it

does, whereas a growing child will be able to tell us. Moreover, the child learns a

language that includes linguistic forms that have no counterpart even in the most

intricate ‘language of the beasts’. A child gradually comes to use words not only to get

what it wants but also to speak about what it is going to do; and comes to understand

and use the locutions by which choices are debated and actions explained, justified, and

recommended. It is the use of such parts of our language, appearing for the most part in

regular ways in the web of actions, that allows us to say, without appealing to a hidden

mental ‘realm’, that ends and means may be ‘in our minds’. It makes sense to ask what

someone thought about the pros and cons of a particular choice because we can ask

him and be given an answer. And he himself can go through arguments that have as a

conclusion ‘so that is what I shall do’. When we say that human beings are able to

choose on a rational ground as no animal can, it is because human action belongs in

such surroundings, and so, ultimately, because humans use language not matched by

anything in animal life. (NG, 55)

This is a kind of Wittgensteinian gloss on the Thomistic principle that humans

distinctively pursue goods qua good. Humans do not simply pursue ends, but they do

so intelligently and articulately. Unlike animals, they can explain, justify, and

recommend actions, and this is an essential part of the human capacity of practical

reason. It is this that formally distinguishes human from animal action. The account

thus remains neutral concerning the sort of reasons agents communicate and act upon.

So Foot’s endorsement of (6) does not require that those reasons be evaluations. It does

not support derivationism.

This is fortunate, for derivationism is too demanding in its understanding of the qua-

good requirement (cf. Nieswandt 2023). The idea that human beings are

characteristically articulate about their reasons does not mean that they are always

ready and able to answer the question of why their reason for acting is what it is—

even, or perhaps especially, if they are virtuous (cp. Williams 1981a, 18). There should

be room in Foot’s account for the idea that reasons “come to an end somewhere,” as

Wittgenstein famously says of explanations (1958, §1). This is not a limitation on what

the virtuous agent knows. Her reasons do not merely to come to an end somewhere;

they come to an end in a situational reason because such is the ultimate ground for

action done virtuously. “The just person aims at keeping promises, paying what is

owed, and defending those whose rights are being violated, so far as such actions are

required by the virtue of justice” (NG, 12).

On Foot’s view, there is a further philosophical question that can be raised,

concerning why some situational consideration is a reason for action. But the answer to

this question is not itself a reason for action within the same order of reasons as the

situational reason. It is appropriate that, if queried, many virtuous people will simply

reject the question of why they are keeping their promises or helping their friends, if

they don’t simply provide some uninformative stock answer (“I couldn’t live with

myself if I didn’t”) (cp. Lott 2015, 84-85). In doing so, they will still be manifesting

their difference from the animals.

2.6 The Normativity of Evaluations

Still, one might think that there is something importantly right about treating

evaluations as reasons for action. It is often thought to be among the principal tasks of

practical philosophy to capture how evaluations are normative, or reason-giving. This

is one way of taking what Christine

Korsgaard says here:

[E]thical standards are normative. They do not merely describe a way in

which we in fact regulate our conduct. They make claims on us; they

command, oblige, recommend, or guide. Or at least, when we invoke

them, we make claims on one another. (1996, 8)

The obliging character of certain evaluations might seem to disappear on my reading of

Foot, and that, it might be said, surely misses something of the phenomenology of

practical thought, as well as of advice.

On the contrary, I want to say that it is precisely wrong to treat evaluations as reasons

for action. The obliging character of evaluations is not best captured by enlisting them

as reasons. This is especially clear in the case of those evaluations which are

themselves reasons statements. My reason for keeping my promise is not that I have

reason to keep my promise; that is hardly intelligible. But is it much different to

propose that my reason for keeping my promise is that I ought to or that I should or

that it would be reasonable to? My reason for keeping my promise is the promise itself

which I made; that reason is why I have a reason to keep it, and also why I ought to,

and why I should, and why it would be reasonable.

This objection to treating evaluations as themselves normative reflects the

phenomenology of virtuous motivation. The virtuous person does not need to be told

that she ought to keep her promise; rather: her promise is already a reason for

performance, the only one she needs. Agents,

indeed, are not primarily in the business of evaluating their own actions; they are rather

thinking about the world and the kind of response it calls for (cf. Ford 2017). “You

ought to do what you promised” is the sort of consideration typically addressed

precisely to someone who is not inclined to do what she promised. And addressed to

such a person, it may well not produce any action.

This is not to say that evaluations have no role to play in the thought of a virtuous

agent.

Consider the following passage of Foot’s:

Every ‘should’-statement that gives even a reason for φ-ing must itself

be based on a certain consideration (or considerations) that can be

quoted in answer to the question ‘Why do you think that is a reason for

φ-ing?’ … Take, for instance, the consideration that a certain man is my

father, as it might stand as evidence for the proposition that there is a

reason for me to look after him in old age. (NG, 58)

Now, Foot here is envisioning that a ‘should’-statement like “I should help this man in

his old age” “gives a reason” for my helping him. What is surely right about that is that

such a ‘should’statement gives a reason (in the sense that and to the extent that we want

to say that) only derivatively from some situational consideration, such as that the man

is my father.

In any case, Foot goes on in passing here to make an interesting remark, which

I wish to highlight. The situational consideration that a certain man is my father, she

says, stands as evidence for the evaluative proposition that there is a reason for me to

look after him in his old age. This is, I want to say, a way in which the concept of a

reason does figure into the deliberative perspective; it is a case where an evaluative

reasons statement is made on the strength of a pertinent situational consideration.

It makes sense that the concept of a reason can figure into the deliberative

perspective in such a way. If I can, as it were, infer the action of helping my father in

his old age from the fact that he is my father, then presumably I can also infer that there

is a reason for me to help him in his old age.

What shall we say is going on here? It seems to me that we should say this: a

subject’s use of a statement such as that p is a reason to φ registers the subject’s

acceptance of a pattern of inference of which “p; therefore I’ll φ” is an instance. A

subject draws on her recognition of the soundness of such a pattern in reflectively

acknowledging that p is a reason to φ. The thought that p is a reason to φ is not

independent of the recognition that the corresponding pattern of inference is sound.

To accept a pattern of inference cannot simply be to make another judgment.

The judgment that p is a reason to φ, after all, states the soundness of the

corresponding inference and does not explain it. Rather, to accept a pattern of inference

is, in the first and most basic instance, to practice it—to φ on the basis that p, say. One

might also accept the pattern in virtue of p’s playing some other role in one’s thought—

for instance, if one would φ on the basis of p, but for the fact that q (in other words, one

treats the fact that p as relevant to the question whether to φ).

That we can make claims about reasons from the deliberative perspective, on

the strength of pertinent situational considerations, in the way Foot’s example

illustrates, seems beyond dispute. The notion of a practical reason (which we saw Foot

deploy above in responding to Watson) is one that human agents get a hold of and

deploy in their own thought and talk. The use

of a practical-reasons statement is licensed not by empirical investigation but by the

recognition of a situational reason and one’s acceptance of a corresponding pattern of

inference.

In making this point, there is nothing special about the notion of a reason in

comparison with other evaluative notions. By drawing upon our own inclinations to

practice certain patterns of inference, we can make judgments from the deliberative

perspective concerning what is reasonable, just, or courageous, what we ought to do,

and so on—if these concepts are part of our vocabulary—though the details of the

appropriate use of each concept will, of course, vary.

Williams makes a parallel point about one sense of ‘ought’:

Whatever other oughts there may be, we can recognise the use of the

expression in the conclusion of deliberation: ‘This is what I ought to do’

expresses the agent’s recognition of the course of action appropriate, all

things considered, to the reasons, motives, and constraints that he sees

as bearing on the situation. (1981b, 124)

The idea then is that one important use of evaluative expressions is to summarize and

abbreviate the results of deliberation and the role of situational reasons within it. When

my considered response to a situation is to φ, I may also say that I ought to φ. Let us

say that, in this way, some evaluations are deliberatively available.

What is the point of this use of evaluative language? It is not to provide a guide to

action that is independent of or additional to the situational reasons one recognizes.

Those already suffice

for a determination of what is the appropriate thing to do and for the doing of it. Rather,

as I have just said, deliberatively achieved evaluations summarize and abbreviate the

tendency of one’s consideration of the reasons available to the agent—or, importantly,

of the reasons available to someone else. To the extent that we can occupy others’

perspectives, we will be able to formulate and address to them evaluations which will

have the force of advice. To say to someone who is in doubt over, say, whether to

attend his sibling’s performance or go out with a friend, “You ought to go to the

performance,” is, as it were, to add: “After all, you promised” or “He’s your brother.”

Such utterances gesture toward situational reasons which they claim to exist, reasons

which the advisee, even if not fully virtuous, may be inclined to recognize and treat as

reasons, if only they were called to his attention (perhaps: if only he is denied the

opportunity to ignore them). It is still the situational reasons which are one’s reasons

for acting, not the evaluations. We might put this point by saying that the normativity

of practical-reasons statements and of evaluations generally is transparent to the

normativity of situational reasons, which is explained through the motivations of the

virtuous agent. This strikes me as an attractive explanation of the normativity of

evaluations.

In any case, the mistake of including evaluations among the virtuous agent’s

reasons is an understandable one, for when the situational reasons speak in favor of a

particular kind of response, the articulate agent will be able to say so in evaluative

terms. The relevant evaluation “gives a reason” for the agent’s response, but not in the

same sense in which a situational reason does.

Evaluations do not have a force that is independent of that of the situational

reasons of which the agent is aware, but nevertheless there is point to saying that an

agent φ-ed because it

was reasonable or refrained from φ-ing because it would have been wrong. This is to

say that there were some situational reasons which grounded the agent’s φ-ing or not φ-

ing.

Now, when I introduced the class of evaluations, I noted that for Foot they are

internally related to natural-historical judgments. That is a fundamental commitment of

neo-Aristotelian naturalism. Can this be maintained, in the case of evaluations that are

deliberatively available? I treat this question in the next chapter. It seems to me that

there is a tension between the deliberative availability of evaluations and Foot’s

particular conception of natural-historical judgments. For, as I will illustrate in Section

3.2, Foot ultimately holds that natural-historical judgments bear an internal relation to

facts about what human beings need. In Section 3.3, I will articulate a doubt whether an

agent could learn such facts on the strength of the consideration of pertinent situational

reasons. But I take this worry to be a problem not for neo-Aristotelian naturalism as

much as it is for Foot’s way of working it out. Foot’s commitment to a need-based

conception of naturalhistorical judgments is a consequence of what I take to be an ill-

advised departure from Thompson’s conception of life (NG, 30-31). On a different,

more Thompsonian conception of the teleology of life, which I spell out in Section 3.7,

one may instead conclude that one learns human natural history through the experience

of virtuous motivation.

2.7 Conducting Oneself, Again

Let us take stock. I have said that Foot’s conception of virtue and of practical

rationality is an attractive one. Some of the reasons for this have already been

mentioned in the course of the argument above. It seems, for instance, to be no more

demanding than is realistic upon the

cognition of ordinary virtuous agents. It captures the fact that a just individual, though

not especially articulate concerning the theoretical foundations of justice, may act

justly all the same.

But it also accommodates nicely the role of practical rationality in conducting oneself

well, which I introduced in Section 1.7. In this section, I take note of a couple relevant

results, concerning the extensibility of Foot’s account to aspects of virtue that are not

so easily accommodated by derivationism.

(a) Not only will the temperate, say, accord appropriate weight to prospective

pleasures, but in many cases they will not notice them at all. Similarly, a just person

does not notice the various ways in which he might advance his own interests by

snubbing others. It does not even occur to him to simply steal what he wants or needs,

and he would be defective as an agent if it did. The wise person will generally be one

who habitually grasps plausible avenues of thought in deliberation, without considering

many possible but unfruitful ones.

What these examples make evident is that virtue includes not just patterns of response

to reasons that are noticed but patterns of noticing. To possess a virtue is be disposed to

notice, and not to notice, various sorts of considerations. This point is not naturally

accommodated on a derivationist account of virtuous agency. For while there is a shred

of plausibility in taking virtuous actions to be conclusions from considerations about

human nature, what one omits even to consider cannot be the conclusion of anything.

This aspect of virtue is, it is true, not explicitly thematized by Foot. But it seems

to be required by the key passage’s characterization of a virtue as a disposition to

respond to certain sorts of reasons. Here is the relevant bit again:

Those who possess these virtues possess them in so far as they recognize

certain considerations (such as the fact of a promise, or of a neighbour’s need)

as powerful, and in many circumstances compelling, reasons for action. They

recognize the reasons, and act on them.

Those reasons that are not noticed cannot be responded to. We need only add to Foot’s

account that virtue is shown also in what is not recognized.

This helps, I think, to account for why it is worth talking about virtue at all.

Korsgaard objects to the key passage in this way:

One might suppose that in granting that what makes a person “just” is

the nature of the reasons on which he acts and the weight he assigns to

them, [Foot] is granting that what makes an agent well-functioning is

his principles. To say that someone is “just” is simply to say that he has

adopted a certain kind of principle: in her words he has “accepted a

certain group of considerations as reasons for action.” But then why

does she want to talk about the virtues? If to say that someone has a

virtue is just to give a third-personal name to the fact that he acts on

certain principles, surely the principles are what moral philosophy

should be concerned with. (2019, 7)

It is true that the fact that φ-ing would be just or temperate or courageous is not

(usually) among the virtuous agent’s reasons for doing it. These judgments largely have

their home in third-personal evaluation; what the agent is thinking about is the world

response to which is necessary. But this does not mean that the notion of virtue is

otiose. A person is not said to be virtuous just because it is her principle to take the

addressable needs of others as reasons for action, for she will only do so if she is

generally the sort of person who notices the addressable needs of others. In one way of

taking the expression, then, having the right principles is not sufficient for conducting

oneself well.

(b) I wish to mention—all too briefly—the relationship between virtue and emotion, a

topic upon which Foot hardly touches. While emotions often do prompt action, they are

also felt and expressed in many other ways. Someone, for instance, cuts me off on the

highway, and I feel a surge of anger; or I realize I have forgotten to keep a

commitment, and I feel a pang of guilt. Quite plausibly, the possession of virtues is not

just a matter of being disposed to act in certain characteristic ways but of being

disposed to feel in certain characteristic ways. The mild person does not feel anger

unless the situation really warrants it, and it will be partly characteristic of the just

person that she feels regret upon recognizing her occasional failure to give to others

what she owes them.

It is not clear that derivationism can do justice to the relationship between

virtue and emotion. It will be possible to ascribe rationales to subjects for their

emotions; we will be able to say, for instance, that I am angry because someone cut me

off and I am in a rush. But in the case of emotion, it is especially implausible that these

rationales will include truths about human nature or evaluations that have such truths

among their truth conditions. Derivationism could, of course, grant this point, but then

it will give an implausibly disunified account of the reasons on which virtuous agents

act and feel, which will—further—be in some tension with the fact that human beings

can act from emotion.

I conclude, then, that there are some important advantages to rejecting

derivationism and the conception of practical rationality which it would underwrite.

Chapter 3: The Teleology of Life

[J]ust as man has so many teeth, which is certainly not the average

number of teeth men have, but is the number of teeth for the species, so

perhaps the species man, regarded not just biologically, but from the

point of view of the activity of thought and choice in regard to the

various departments of life—powers and faculties and use of things

needed—‘has’ such-and-such virtues. (Anscombe 2006b, 188)

3.1 Introduction

In Chapter 2, I defended a sympathetic reading of the main thrust of Philippa Foot’s

Natural Goodness (NG). Foot, I argued, is not a derivationist. She does not hold that

facts about human nature are among the virtuous agent’s reasons for acting well. The

virtuous agent, rather, acts on the basis of characteristic sorts of what I termed

situational reasons—“I promised to φ,” “X is hers,” “Y is dangerous,” and so on. The

role of human nature in the argument is different than it would be in a derivationist

account; the dispositions to act for the appropriate classes of situational reasons are

virtues (and the agents who possess them are practically rational) because they are

dispositions that human beings need to have, if they are to achieve their good.

I mentioned but did not treat this last point, concerning human nature and its relation

to human need. I want in this chapter to dwell upon it, for it is of considerable

importance in understanding Foot’s conception of life and its application to human

beings. It is an instance of a more general thesis of Foot’s concerning the constitution

of an organism’s natural history: (1) A feature F is part of the natural history of Ss only

if it is an Aristotelian necessity, for the attainment of the good of Ss, that a given S be

To say that a feature F is part of the natural history of Ss is equivalent to saying that “Ss

are F” is a true natural-historical judgment, so (1) is effectively a claim about the truth

conditions of the natural-historical judgment “Ss are F”.

(1) refers to so-called Aristotelian necessity. Foot adopts this notion from G. E. M.

Anscombe’s work on rules, rights, and promises. Much more will be said about it

below. For now, suffice it to say that X is an Aristotelian necessity for Y if Y is a good

the realization of which depends on X. Aristotelian necessity, then, is a species of

teleology; X’s being an Aristotelian necessity for Y is one sense in which X may be

said to be for the sake of Y. So (1) is, effectively, a stance on the kind, or at least a kind,

of teleology which characterizes life.

This chapter is an argument against invoking Aristotelian necessity, as Foot does, in an

account of the teleology of life. I set out Foot’s account and take note of the pressures

within it that push toward the adoption of (1). I argue that the account faces

considerable difficulties, which are thrown into relief by Foot’s attempt to apply it to

human beings; I argue also that (1) is in tension with what I have called the deliberative

availability of evaluations. Finally, I propose an alternative account that meets Foot’s

desiderata while leaving possible, consistently with neo-Aristotelian naturalism, the

deliberative availability of evaluations. The alternative account rejects the role Foot

assigns to Aristotelian necessity, putting in its place a sui generis form of teleological

relation while also reconceiving the relationship among the various elements of a

natural history.

3.2 Aristotelian Necessity

Before we consider the question of whether Foot’s account is satisfactory—and if not,

what might replace it—we need to get more of it in view. Does Foot actually hold (1)?

And what does the notion of Aristotelian necessity come to? I address these questions

in this section.

Foot aims, following Thompson, to provide an account of the logical form of

naturalhistorical judgments, since the thesis she would like to defend—the thesis I

labeled neoAristotelian naturalism in the previous chapter—is that that very form is

shared by moral judgments. As we saw there, she rehearses Thompson’s many

arguments concerning the distinctive, “non-Fregean” sort of generality exhibited by

natural-historical judgments. But she diverges from his account at a crucial juncture.

Natural-historical judgments are a species of generic sentence. That is just what

grammarians call general sentences of such forms as “Ss are F,” “The S is F,” and “An

S is F,” in which a predicate is attached to a bare plural or (in)definite singular term.

Not all generic propositions are natural-historical judgments. According to Foot,

Thompson does not give us the resources to distinguish two superficially similar sorts

of generic propositions about living things, exemplified by the following:

(2) The blue tit has a round blue patch on its head.

(3) The male peacock has a brightly colored tail.

The color of a male peacock’s tail serves a purpose in its mating and thereby in the

reproduction of peacocks. If it were the case, though, that the blue tit’s blue patch

played no such role, then it would not be appropriate, Foot thinks, to say that a

particular blue tit was defective in lacking a blue patch. It is therefore necessary to

distinguish “the teleological from the non-teleological attachment of predicates to a

subject term that is the name of a species” (NG, 30). Natural-historical judgments are

those in which the attachment of the predicate is “teleological.”

We could say … that part of what distinguishes an Aristotelian

categorical from a mere statistical proposition about some or most or all

the members of a kind of living thing is the fact that it relates to the

teleology of the species. It speaks, directly or indirectly, about the way

life functions such as eating and growing and defending itself come

about in a species of a certain conformation, belonging in a certain kind

of habitat. This is why the noise made by the rustling of leaves is

irrelevant in this context while the development of roots is not. And this

is why Aristotelian categoricals are able to describe norms rather than

statistical normalities. It matters in the reproductive life of the peacock

that the tail should be brightly coloured, whereas our assumption has

been that the blue on the head of the blue tit plays no part in what here

counts as ‘its life’. And this is why the absence of one would itself be a

defect in an individual whereas that of the other would not. (NG, 33)

Some care should be taken here. Foot writes as though, if a generalization is not a

natural-historical judgment, then it is a mere “statistical normality” (cp. also Lott

2018). This is not the best way of putting matters, since even the wider class of generic

sentences exhibit non-Fregean generality. Consider, for instance, the statements that

mosquitoes carry the West Nile virus or that sharks attack bathers, which are generics

but are not natural-historical judgments (Leslie 2015). Granting that point, though, we

can find plausible the nearby, Foot-friendly suggestion that what distinguishes natural-

historical judgments from the wider class of generics is their teleology. This is a

suggestion to which I will return in Section 3.7.

Now, in the case of a non-human species S, Foot says that a feature F, if it is to be

relevant to the natural goodness of Ss, must relate in some way to the “life cycle” of

Ss: to their development, self-maintenance, and reproduction (NG, 41). What exactly is

this relationship of teleology between the naturally good features of an organism and

its life cycle? Foot speaks variously of the relation in question. It is a relation of

“playing a part in the life,” as “having to do … with self-maintenance … or with the

reproduction of the individual,” as “being causally and teleologically related to [the

life],” as “serving a purpose,” as “giving the ‘how’ of what happens in the life cycle,”

and as being “determined by what is needed for development, self-maintenance, and

reproduction” (NG, 31-33). Her considered, official view, is that the relation is that of

Aristotelian necessity, “that which is necessary because and in so far as good hangs on

it” (NG, 15). As I have said, Foot borrows this notion from Anscombe’s work on

promising, rights, and political authority, in which Anscombe cites Aristotle’s remark

that in one sense ‘necessary’ means that without which some good cannot be

(Anscombe 1969, 73-75; 1981b, 97-103). To get this idea in view, it will help to take a

closer look at the use Anscombe makes of it (cf. also Vogler 2020). I will focus on the

case of promises.

Anscombe observes that promises are a kind of linguistic instrument. Their

characteristic way of operating is in outline like this: if someone has made a promise,

then she has to keep it, and in keeping it, she may cite as her reason for performance

the fact that she made the promise. This is the “language-game” that is played with the

words “I promise …”. But to say only this is not to distinguish promise-making and -

keeping from the playing of a mere game. A chess player, we may say, “has” to move

her bishops diagonally, or may “have” to move her king out of check, or “has” to begin

by moving either a pawn or a knight. But she only “has” to do these things if she means

to be playing chess. There is nothing wrong with walking away. Besides being a bad

sport, there may not be anything wrong with breaking the rules or upending the board.

We can even imagine, as Anscombe does, a kind of game in which one gives a

description of an action,

accompanied by some special sign, and thereby “has” to do it. One might prefix action

descriptions with ‘Bump!’ or write them in green ink, and then one might complain in

the game against someone who does not, say, φ: “But you bumped to φ!” or “You

green-inked it!”

Such a game would be similar in its workings to the institution of promising.

But the objection to someone who does not play this game would be no more than that

she is not doing so, whereas the objection to someone who does not keep her promises

is more than this: she is unfaithful, and dishonest too, if it had been a lying promise.

She has behaved wrongly, badly. Anscombe’s question is: How is this so? How can the

keeping of promises be an ethical matter? By making a promise, I (typically) make it

the case that I will act badly in failing to do something which I did not otherwise, in

any sense, have to do. How do I come thus to restrict my future possibilities for acting

well, just by uttering the words “I promise …”? How is it that uttering these words

creates a special kind of incumbency upon me? As Anscombe memorably puts it, what

needs to be shown is why “You promised but did not do it” is “a reproach of any more

significance than

‘You eat your peas with your knife’. One may ‘go along with’ such a reproach to a high

degree …—But why not break the spell?” (Anscombe 1969, 72).

The notion of a language-game has been brought in to account for the inner

workings of the practice of promising, but Anscombe thinks that that is where its

philosophical utility ends. The notion is insufficient to answer this question we have

just encountered, of how promising restricts my future possibilities of acting well. It is

at this point that Anscombe appeals to Aristotelian necessity. A great deal of good

depends on the institution of promising, for it is a tool by which human beings can get

others to do what they need done, and the alternative means for addressing this basic

and pervasive need are extremely limited (Anscombe 1969, 73-75). In other words, the

necessity of keeping promises lies in the fact that humans need the institution of

promising. It is, in Foot’s phrase, an institution on which good hangs. Such an account

of promising, if it works, secures that one generally acts well in keeping one’s promises

and badly in failing to do so, even when (as occasionally happens) there are not great

goods at stake in the keeping of a particular promise.

It is this notion which Foot takes to relate the naturally good characteristics and

operations of an organism to its life cycle. “We invoke the same idea when we say that

it is necessary for plants to have water, for birds to build nests, for wolves to hunt in

packs, and for lionesses to teach their cubs to kill” (NG, 15; cf. 31-33).

3.3 A First Worry

This sketch suffices to raise my first worry concerning the role of Aristotelian necessity

in Foot’s argument. Before continuing my exposition of the relevant parts of Foot’s

account, I will develop this worry in this section. My worry bears upon the

compatibility of Foot’s account with what I called (in Section 2.6) the deliberative

availability of evaluations.

I suggested there that, in one of its central uses, evaluative language abbreviates and

summarizes the weight of what one takes to be one’s situational reasons. For instance,

suppose that I φ on the basis of the fact that p. It seems, and Foot appears to agree, that

on the strength of that very situational reason, I can also infer that I have a reason to φ,

namely that p. But

(4) p is a reason to φ is an evaluation of human practical reason. Evaluations of

human practical reason, are on Foot’s neo-Aristotelian naturalism, judgments of natural

goodness or defect. And judgments of natural

goodness and defect are internally related to natural-historical judgments. The latter are

among the truth conditions of the former. That point is central to the conception of

natural goodness and defect that Foot and Thompson share.

Now suppose too what (1) implies: that facts about human need are part of the

truth conditions of natural-historical judgments. By transitivity it follows that facts

about human need are part of the truth conditions of evaluations such as (4). If it is

true, as the early Wittgenstein says, that “[t]o understand a proposition means to know

what is the case if it is true” (1974, 4.024)—if it is true that to understand a proposition

means to know what follows from it, to know what its truth means for the world—then

it seems that the deliberative availability of evaluations implies the deliberative

availability of facts about human need. Given (1), I do not genuinely understand the

evaluation of human practical reason which I make, unless I understand also its

connection to human need. So if I can infer an evaluation of human practical reason on

the strength of the situational considerations that are pertinent in a particular context,

then I can likewise infer facts about human need.

But—this is the objection—that is incredible. Take the general facts of human life to

which Anscombe appeals in her account of promising; or take the fact, to which Geach

alludes, that “any large-scale worthy enterprise” will require some measure of courage

and temperance (1977, 16). Are these frankly rather sophisticated facts knowable on

the strength of pertinent situational reasons, such as that I made a promise or that such-

and-such is dangerous? I submit that they are not. The fact that certain patterns of

behavior are necessary for human life, constituted as it is, to go smoothly, is not the

sort of thing one can plausibly be said to know in virtue of one’s capacity to recognize

the class of situational reasons which characterize one or another virtue.

The argument for this conclusion has, of course, multiple premises, and its incredible

conclusion can be avoided by abandoning any one of them. One could, for instance,

deny that evaluations are deliberatively available, though in my view that does too

much violence to our evaluative thought and talk. (Derivationism is a way of denying

the deliberative availability of evaluations. But I have argued that derivationism is not

an attractive view.) Alternatively, one could give up neo-Aristotelian naturalism, and

hold that in making evaluations we do not tacitly commit ourselves to anything’s being

the case about human nature. Foot will not want to say that. What I want to suggest is

that neither reaction is necessary. Instead, (1) can be given up. Indeed, there are

independent reasons for doing so. It is in that direction that I now turn.

3.4 Foot on Human Good: I

(1) says that a feature of an organism, if it is to be part of the species’ natural history,

must be an Aristotelian necessity for the organism’s good. In light of the foregoing, this

might seem odd, for we have seen that Foot takes the natural-historical features of non-

human organisms to be, specifically, for the sake of their life cycle—their development,

self-maintenance, and reproduction.

Foot eventually looks back at her treatment of non-human life and sees an even more

general notion than that of the life cycle at work. “[I]n the course of describing ‘natural

goodness’ in plants and animals,” Foot says, “we have implicitly adverted to the idea of

the good of a living thing as well as its goodness in various respects, and the two ideas,

though related, are distinct” (NG, 41). For Foot, then, it is important to distinguish an

organism’s goodness—that is, of features in virtue of which it is naturally good in

particular respects—from its good. In non-human creatures, the good is said to be the

life cycle of development, self-maintenance, and reproduction.

“But clearly,” Foot says, “this is not true when we come to human beings” (NG, 42).

Human beings are distinctive not only because human life is more complicated

but because the human good is genuinely richer than the life cycle. It is not just that

survival and reproduction take on a different significance in human life, though they

do. Survival and reproduction are also less dominating and central, because distinct

goods also come onto the scene (NG, 43). Various examples are given throughout the

book. The enjoyment of stories, songs, dances, and jokes, Foot says, fall under human

good, and this is indicated by our sense that humans who lack the specifically human

powers required for such enjoyment are “deprived,” even if they can survive and

reproduce (NG, 43). Her comments on the ends that some of the virtues serve indicate

that she thinks human good also includes “love and friendship” (NG, 44). Later, Foot

denies that the sort of felt contentment sometimes called “happiness” is part of human

good (NG, 83-85). Yet, she argues, there is a sort of happiness, “deep happiness,” with

which the human good can even be identified. Deep happiness is supposed to be

incompatible with wickedness and is said to involve “home, and family, and work, and

friendship,” “the quest for truth,” “artistic creation,” and “the exploration of strange

lands” (NG, 86-97)

Human good includes more than the life cycle, more than development, self-

maintenance, and reproduction. To put the thesis in Foot’s terms:

(5) Human good is sui generis. (NG, 51)

What role does (5) play in Foot’s argument? What Foot would like to establish, in order

to secure neo-Aristotelian naturalism, is the view that practical rationality (the good

disposition of human practical reason) involves the possession and exercise of the

virtues. That is to say, she would like to establish that humans are naturally good in

respect of their practical reason to the extent that they possess and exercise the virtues.

Given her amendment to Thompson, establishing this requires arguing that the

possession and exercise of the virtues is an Aristotelian necessity for human beings—

that their good hangs on it. And this is indeed what we find her arguing:

[F]or all the diversities of human life, it is possible to give some quite

general account of human necessities, that is, of what is quite generally

needed for human good, if only by starting from the negative idea of

human deprivation. For then we see at once that human good depends

on many characteristics and capacities that are not needed even by

animals, never mind by plants. (NG, 43)

Men and women need to be industrious and tenacious of purpose not

only so as to be able to house, clothe, and feed themselves, but also to

pursue human ends having to do with love and friendship. They need

the ability to form family ties, friendships, and special relations with

neighbours. They also need codes of conduct. And how could they have

all these things without virtues such as loyalty, fairness, kindness, and

in certain circumstances obedience? (NG, 44-45)

Foot’s claim is that the virtues are among what is needed for the attainment of human

good. They are among the Aristotelian necessities of human life. She takes this to

establish that they are naturally good and that to possess them is to be practically

rational.

(5) will have an intelligible, if not irresistible, appeal to anyone making this

argument. If each natural-historical judgment about human beings could only be

referred to human development, self-maintenance, and reproduction, Foot thinks, then

there would be reasonable grounds for doubt whether the account issues in anything

like the traditional catalogue of the virtues, that is, whether its morality is recognizably

ours. The account would appear objectionably biologistic. I do not mean to understate

the no doubt real degree to which even ‘mere’ development, self-maintenance, and

reproduction among humans would require a textured and at least somewhat familiarly

human kind of life. Humans indeed could achieve nothing if they did not live together

and cooperate in some significant measure, and this could not be achieved without a

great deal of trust, truthfulness, and justice; and they would be substantially hindered in

many projects if, lacking courage and temperance, they could not face dangers and

resist temptations of various kinds. Much could be (and has been) said on this front, to

be sure. But Foot herself concedes that human good needs to be extended beyond the

life cycle if her account is to yield a recognizable morality. That is why (5) is so central

to her attempt to apply the framework of natural goodness to human beings—she

devotes a chapter to articulating and defending it. And that is also why the argument for

the natural goodness of the virtues, which we have just seen quoted above, does not

come until we have seen (5) defended.

Foot’s acceptance of (5) means that she denies that the form of natural-historical

judgments makes essential reference to the life cycle. For natural-historical judgments

about human beings make essential reference to something more. Nevertheless, Foot

insists that as we move from the non-human case to the human one, “the conceptual

structure [of natural goodness] seems to be intact,” for human good, though enriched in

comparison with non-human good, continues to play the same role in the evaluation of

human characteristics and operations as “flourishing” plays in the evaluation of non-

human ones (NG, 44). That is, just as the good features of non-human organisms are

Aristotelian necessities for their good (their life cycles), so the good features of humans

are for the sake of their good.

3.5 Foot on Human Good: II

On what grounds, exactly, does Foot hold (5)? It seems to me that the raising and

answering of this question tends to reveal substantial problems for Foot’s account. In

this section, I will consider her answer to it, which will lead us to recognize some

constraints Foot places on an adequate account of life. In the next section, I will argue

that Foot’s way of satisfying these constraints is objectionable.

Foot does have a kind of method for identifying the good of a species and thus

for judging that human good is sui generis. At various points, she adverts to the notion

of a benefit to argue that something is or is not part of the good of a species. The initial

reason for introducing this notion is to argue that it is necessary to distinguish the good

of a living thing from its natural goodness, on the basis that these two can come apart

(NG, 41-42, cf. 34). What is naturally good for an organism is what makes it a good

member of its species—good qua member of its species. On Foot’s account, an

organism that is good of its kind is an organism that has such characteristics and

engages in such operations as are necessary for its good. But an organism that is

naturally good in this sense can still, intuitively, fail to attain its good, as in the case of

a bee that dies when it stings what threatens the hive. The bee was a naturally good bee

and it behaved well, but intuitively it has not been benefited—as Foot wants to say also

in the case of someone who possesses courage unto death, or justice unto penury (NG,

43; cf. 35; following Geach 1977, 17). A naturally good organism can even behave in

such a way as leads to the destruction of the group.

For example, a bee might sting a gardener who then destroys its nest.

Foot is consistent in appealing to this notion of benefit as a criterion—a mark by

which to judge—for determining what the good of a species is. It, along with the

converse notion of deprivation, is her criterion for holding that human good is sui

generis; as we have seen, Foot remarks that without, e.g., the telling of stories and

jokes, human life would be “deprived” and “impoverished” (NG, 43). In other words,

she registers an intuition that non-human life is only benefited by what serves its life

cycle, while humans are benefited by various other goods. Similar intuitions are

employed in her treatment of happiness, to rule out mere contentment or satisfaction as

a genuine human good and to argue that deep happiness, by contrast, is part of, or even

identical to, human good. We would not say, she thinks, that we have benefited

someone if we have helped him in the perpetration of grave vice, and so some kind of

happiness incompatible with wickedness seems to be the human good (NG, 84-86, 93-

94).

Details remain about how such a procedure is to be executed. No doubt it raises

various questions, but this does in outline seem to me to be Foot’s account. For Foot,

the features that make an organism naturally good are those that it needs to have if it is

to attain its good, which Foot identifies by way of our sense of what benefits the

organism.

Some readers of Foot will disagree with this reading because of the prominence

it gives to the notion of benefit—for the appeal to benefit, on my telling, is effectively

an appeal to intuition, and accordingly it is an odd thing to find in the writing of an

author who was so critical of G. E.

Moore’s intuitionist account of the good (Moore 2004; NG, 2-3).

It is indeed a strange thing to find in Foot’s writing. But, I have argued, it is there. And

while Foot seems to me to have been insufficiently reflective about what her appeal to

benefit involves, it plays an central role in her argument; we can see this if we attend to

the shortcomings of an alternative reading, provided by Micah Lott, which—I think—

would not serve all of Foot’s dialectical purposes if it represented her view.

Lott argues that the good of an organism is its “good life” and glosses this in

the following way:

(6) “THE GOOD OF any creature is the actualization of those well-formed

capacities that make it GOOD AS [that is, naturally good as] the kind of creature that it

is” (2018, 258). Lott thus employs a distinction between what Aristotle calls first and

second actuality. Aristotle distinguishes two different kinds of transition from

potentiality to actuality. There is the acquisition of the capacity, and then there is the

exercise of the capacity possessed (1984a, 417a21-b1). Some examples will make the

idea clear:

First potentiality

First actuality /

Second potentiality

Second actuality

Example The capacity to learn a

language

The capacity to speak Speaking

Example The capacity to become a

fast runner

The capacity to run fast Running fast

Example The capacity to acquire

virtue

The capacity to act from

virtue

Virtuous activity

Table 2: Examples of first and second actuality

Lott uses this distinction to provide a reading of Foot’s various uses of ‘good’. On his

reading, an organism is naturally good insofar as it possesses the first actualities that it

is supposed to—those determined by its life form. It achieves its good when those first

actualities, or perhaps certain important ones, are themselves actualized.

It is not clear to what extent this is meant to be a reading of Foot. Lott clearly does

intend for his account to shed some light on her various uses of ‘good’. But Foot never

discusses the distinction between first and second actuality. Further, Aristotelian

necessity—which is clearly

central to Foot’s conception of the relationship between natural goodness and the good

—is absent from Lott’s account. Indeed, Lott is elsewhere (2015) critical of the role of

Aristotelian necessity in Foot’s argument. Lott’s account is perhaps better construed as

a congenial amendment to Foot’s views.

The distinction between first and second actuality, however, will not quite do all of the

work that Foot wants the distinction between natural goodness and a thing’s good to

do. It handles the cases of the stinging bee and the swift deer, in which what is

undeniably a naturally good specimen does not attain its good; on Lott’s view, this is

because the organism has found itself in less than optimal circumstances in which its

natural goodness in one respect happens to thwart its natural goodness in others (2018,

264). The swift deer is naturally good, good qua deer, in respect of its speed, but in its

present circumstances in which there is a trap this happens to lead to its premature

death, and this prevents the actualization of its other naturally good capacities over a

whole life—that is, prevents it from living the sort of life that it is the good of a deer to

live. But Lott’s account does not handle another class of examples important to Foot.

In particular, Foot’s distinction between natural goodness and the good of a species is

incipient already in the modification of Thompson’s view, described above. As we saw,

Foot contrasts the hypothetical blue tit’s blue patch with the male peacock’s colorful

tail. Similarly, Foot asks whether dancing (as a kind of signal to the hive) is natural to a

certain type of bee (NG, 35, 109). The point is that however typical a feature of an

organism appears to be, it is not to be judged defective for lacking it unless the feature

is actually necessary for the good of those organisms (NG, 30). In both of these cases,

it is by adverting to the good of the species, the life cycle, that we can determine

whether the possession of some feature really redounds to an organism’s natural

goodness. As Foot says, the naturally good features of non-human organisms “are to be

identified by reference to … ‘biological’ cycles” (NG, 41). In other words, Foot takes

the good of a species to serve an epistemic function, giving us a principle by which to

sort out accidental from essential features of species.

Foot is, evidently, interested in prohibiting this hasty inference:

These bees dance.

This bee does not dance.

So this bee is defective for not dancing.

For Foot, this will be sound only if the first premise is a genuine natural-historical

judgment in which the attachment of the predicate is “teleological.” It must be that

dancing is needed for the good of these bees, if such an inference is to be sound. It

would be inapposite to say that dancing is needed for the good of these bees because

(as you can see) these bees dance, and so it is in their nature to dance, and so the good

life of these bees includes dancing. This would allow inferences of the form Foot

means to block. (In blocking them, no doubt Foot has in mind such evaluatively

irrelevant differences among humans as skin color.)

So we have identified certain desiderata Foot would like her account of the teleology

of life to meet. In addition to (5), such an account must preserve the following:

(7) There is a difference between an organism of species S’s being good qua S

and its achieving the good for Ss; there is a difference between its doing

well and its faring well.

(8) The teleology of an organism’s natural features plays an epistemic role in

our distinguishing genuine from ingenuine natural-historical judgments.

Foot is pushed to (1) as a way of meeting them. Lott’s account of Foot’s use of

‘good’—(6)—is tailored to accommodate (7), but it does not on its own help to explain

the truth of (8).

3.6 A Second Objection

Foot argues for (5) by appeal to our sense of what benefits human beings. Moreover,

per (8), the sui generis account of human good is itself indispensable for providing an

account of human natural history, in particular for including the traditional virtues in

that account. These commitments betray a fundamental tension in Foot’s account.

How do we know that human beings are benefited by what conduces to the sui

generis human good? It is obvious that part of our ground for making that judgment

must lie in our grasp of the fact that human beings are different from non-human

organisms. Our natural history is different. The problem is that if (8) is true, then it is

hard to see how we can appeal to that fact in understanding the sui generis character of

human good. For our judgment that human good is sui generis depends on our already

possessing an understanding of human natural history, and our understanding of human

natural history depends on our judgment that human good is sui generis. We are caught

in a circle, which seems indeed to be vicious.

One might suggest that this objection can be answered by way of the so-called

Neurathian procedure of gradual modification endorsed by many contemporary neo-

Aristotelian ethicists (McDowell 2001b; Hursthouse 2001; Lott 2015). Along these

lines, one might suggest that the interdependence of human good and human nature

does not pose a problem for our coming to at least an approximate knowledge of

human nature, because one’s natural-historical judgments need only begin with a

partial, perhaps strictly incorrect, reference to human good. One can refine one’s

conception of human nature and of human good simultaneously to approach the true

natural history of human beings.

But the Neurathian procedure will not help the conception presently under

consideration. To see this, consider two candidate natural histories of the human being.

In one, each naturalhistorical judgment makes reference only to the life cycle G of

humans; in the other, each naturalhistorical judgment makes reference to Foot’s sui

generis human good G'. What is it to judge that one of these natural histories is wrong

and the other right? Foot’s account suggests that one would have to judge that what

conduces to what is contained in G' over and above G benefits humans. On what,

though, could such a judgment be based? It cannot depend on a judgment about the

natural history of humans, for the first natural history would suggest that what is

contained in G' over and above G does not benefit humans. Neurathian modification

cannot begin, because there are no criteria available for the soundness of individual

steps.

Where the good of a species is stipulated, as it were, in advance, as it might appear to

be in the case of non-human creatures, the problem will not arise. A disagreement over

whether it is naturally good for a blue tit to have a blue patch (that is, a disagreement

over the nature of blue tits) is settled by asking whether it is necessary for the blue tit’s

life cycle. But if such a disagreement were traced to a disagreement over what is

actually good for that species—whether the good for blue tits really just is their life

cycle, or whether it is something more—then there is in principle no way to resolve it.

To say that it is a disagreement over what benefits blue tits is really just to redescribe

the disagreement, not to give a procedure for resolving it—unless the procedure is to be

recourse to intuition, about which we would then have the question: what is it intuition

of?

I said that the problem does not appear to arise in the case of non-human creatures.

That is because in the case of non-human creatures, it only appears to be the case that

the good, the life cycle, is intelligible in advance of understanding the natural history of

the organism in question. For even non-human species do not actually all have the

‘same’ good, ‘the life cycle’. The life cycle of blue tits is not the same as the life cycle

of peacocks or oak trees or humans. If the life cycle is self-maintenance, growth, and

reproduction, then its description tacitly involves a representation of the unity of the

organism. A peachick’s end is to become a mature peacock, to maintain itself as such,

and to produce more peacocks. This suggests that the problem lies not so much in

Foot’s attempt to defend (5), which concerns human beings in particular, but in her

account of life itself.

3.7 An Alternative Account of Teleology

Foot opens herself up to the argument of the previous section because she appeals to

the notion of benefit to meet her desiderata (5), (7), and (8). The notion is introduced to

make sense of (7)’s distinction between doing well and faring well. Once it is in hand,

it seems that one can appeal to benefit to specify in what the good of some sort of

organism consists, compatibly with its good extending beyond its life cycle, as (5) says

is the case with human beings. And once the good is specified, it would seem that one

can characterize the natural history of that sort of organism in terms of what is

necessary for its good—and thus teleology will help one sort genuine from ingenuine

natural-historical judgments, as (8) requires.

In this way of proceeding, the good of an organism is specified in advance of its

natural history. We then learn its natural history by referring its features to its good.

The argument of the previous section simply exploited a tension involved in regarding

this procedure as possible.

The model of Aristotelian necessity, though, does seem to encourage the idea

that the good of an organism should be specified in advance of its natural history. For

in Anscombe’s argument, the goods which the institution of promising subserves are

specifiable independently of that institution. They must be, if the argument is to work.

It would make little sense to argue that the institution of promising is necessary on the

basis that, without it, the great good of fidelity cannot be achieved. That is true, but

fidelity is only good because promise-keeping is.

This makes Aristotelian necessity a poor model for biological teleology. If the

good of an organism is, as Foot says, its good life, then this is not a state specifiable

independently of the naturally good features of the organism—because what counts as

a good life for an S depends on what Ss are like. We should find an alternative picture

of biological teleology that accommodates this point, while doing justice to Foot’s

desiderata.

Contrary to the impression Foot gives, Thompson does discuss the relationship

between natural-historical judgments and teleology—though the discussion is modified

somewhat in his book Life and Action, published after NG. Thompson writes:

We are [in the case of biological teleology] … as far as can possibly be

imagined from the category of intention or psychical teleology—a fact

that is also shown in this, that if the complex thought about, for

example, the blossoms is true, then the judgments joined in it are also

true. Here, that is, “P in order that Q” straightforwardly entails both P

and Q. In making out this sort of connection one links a plain fact, not

with a possibly unrealized end, but with another plain fact.

(Thompson 2008, 78)

Thompson is here concerned to emphasize a formal difference between ascriptions of

biological and of intentional teleology. In the course of making this point, he claims

that the relevant ‘in order that’ relation links one natural-historical judgment to

another. Foot herself makes this point: There is an Aristotelian categorical about the

species peacock to the effect that the male peacock displays his brilliant tail in order to

attract a female during the mating season. … Plants grow upwards in order to get to the

light, [though] it is fanciful to say that that is what my honeysuckle is trying to do or

that that is ‘its end’. (NG,

31)

But Foot’s and Thompson’s accounts of this teleology ultimately diverge.

Thompson goes on: Natural teleological judgments may thus be said to

organize the elements of a natural history; they articulate the relations of

dependence among the various elements and aspects and phases of a given

kind of life. I said above that the conditions required for some natural-

historically attributable phenomenon to arise will themselves be natural-

historically attributable to the life-form in question and belong to the natural

history of the form. If these conditions include a feature of the environment,

still the system will contain the judgment that they live in such an

environment, and then in turn judgments about how they come to be there

—and so we go on, never leaving the system of natural-historical

judgments. The teleological connective simply expresses the concept that is

converse to this conception of dependence. (Thompson 2008, 78-79)

If P in order that Q, then P and Q are both true natural-historical judgments, and Q

itself is teleologically articulable. Hence Thompson suggests that a natural-historical

judgment might be uniquely identified—that is, to the exclusion of other generic

propositions—as a “teleologically articulable non-Fregean generality” (2008, 79).

Thompson’s teleological connections thus form what Matthias Haase calls “a closed

cycle” (2018, 93-94), for as he says above, as we move from item to item up or down

the teleological chain, “we go on, never leaving the system of naturalhistorical

judgments.”

Isn’t this a component of Foot’s view as well? After all, she speaks of life cycles. But

while we should certainly ascribe to Foot the best view that we can, we cannot ascribe

to her the idea that natural-historical judgments form a closed cycle. It is not a point

she thematizes or makes use of, even to distinguish organisms from artifacts, as she is

keen to do. There is, moreover, no trace of the thought in her transition to discussing

the sui generis human good. Finally, she simply seems to deny it:

What conceptually determines goodness in a feature or operation is the relation,

for the species, of that feature or operation to survival and reproduction, because it is in

that that good lies in the botanical and zoological worlds. At that point questions of

‘How?’ and ‘Why?’ and ‘What for?’ come to an end. (NG, 42, emphasis added)

Evidently, Foot speaks of the good of non-human organisms as the life cycle not

because she takes cyclic teleology to characterize life and to distinguish it from artifice,

but simply because development, self-maintenance, and reproduction form a cycle.

We can appeal to the teleology of natural-historical judgments—their being linked

with other natural-historical judgments in a “closed cycle”—to accommodate (7), that

is, to address Foot’s problem of the blue tit and the rustling of leaves. Genuine natural-

historical judgments are distinguished from mere generic propositions by their role in

such a teleological system, not by their being necessary for an independently specified

good.

There is no suggestion in Thompson’s account that the teleological connections

between natural-historical judgments are connections of Aristotelian necessity. There

would be something puzzling about that if it were true. For it is not clear that necessity

is the appropriate idea for capturing the way in which life gets on. There is a great

variety of life, after all, and it accomplishes its ends by a variety of means. Hunting is

not the only way of securing nourishment. It is how wolves do secure it. Is it how they

must? Well, considering how they are constituted—perhaps. But how they are

constituted is a matter of their natural history; it cannot be invoked in an explanation of

what their natural history is.

Foot’s appeal to Aristotelian necessity strikes me as basically reductive. It is an

attempt to reduce biological teleology to some other more familiar and tractable kind. I

am not persuaded that this is possible. It may be, alternatively, that the particular

determination of the teleological connective ‘in order that’ which characterizes natural-

historical judgments is sui generis. It is something that we all recognize and cotton

onto from, as it were, our life with the concept ‘life’, from our comprehension of the

proverbial nature documentary. The comparatively non-reductive character of

Thompson’s treatment of teleology may explain why Foot saw fit to amend it, but I

want to suggest that we should not rush to assume that an analysis of biological

teleology is requisite and possible.

What Thompson may seem to have a harder time accommodating is (6)—the

distinction

between doing well and faring well. For that which the natural-historically attributable

features of organisms are for the sake of is now not something categorically different,

as it sometimes appears to be in Foot’s account, but rather: the natural-historically

attributable features of a given kind of organisms are for the sake of other natural-

historically attributable features of those organisms.

But now that we have an account of natural history that is independent of what Foot

calls the good, we are free to give an account of the good in terms of natural history, as

Lott does. We may say that the good of an organism S is its good life; it lies in the

actualization of its naturally good capacities over a whole life. This means that our

account of life needs to appeal to the notions of capacity and actualization, but that

should not be, I think, particularly surprising.

Space is also made for (5)—the idea that human good is sui generis—at least on a

certain way of understanding that thesis. It is, on the view being developed, a

distinctive kind of teleological unity that characterizes life of all sorts. Something is

living if its features can be described—that is, are linked together—in the manner

characteristic of life; this is a formal characterization of what it is for something to be

alive. It is compatible with considerable material variation, from species to species,

concerning what is unified. It will turn out to be a datum, I think, that human life is not

a mere struggle for survival—that humans tell stories and jokes, create art and play

sports, form friendships, theorize, and so on.

Mere satisfaction of Foot’s desiderata does not open up an account of the teleology of

life to the argument I developed in Section 3.6. The problem arises for Foot’s account

as a result of her appeal to the notion of benefit—more generally, as a result of her

committing herself to taking the good of a species to be specifiable in advance of the

species’ natural history. On the present account, if we say that human life is

impoverished without, e.g., artistic expression, then this is best regarded as a kind of

heuristic device that elicits what we already know about the human life form, rather

than a privileged manner of determining what human good is. That is to say, the

account we are constructing rejects the special role Foot appears to give to the notion

of benefit, and more particularly it rejects also the view that human good could be

specified in advance of arguing that the virtues are part of human natural history.

3.8 Nature, Upbringing, and Moral Knowledge

The last problem we are left with is the compatibility of neo-Aristotelian naturalism

with the deliberative availability of evaluations. I introduced this idea in Section 2.7

and raised a worry about its compatibility with Foot’s brand of neo-Aristotelian

naturalism in Section 3.3.

Simply by denying (1), we render neo-Aristotelian naturalism compatible with

the deliberative availability of evaluations, for we avoid the incredible conclusion that

we can learn facts about what humans need by consideration of pertinent situational

considerations, if we deny that human natural history is constituted by facts about what

human beings need. Still, if we are neo-Aristotelian naturalists, we are committed to

the deliberative availability of some human natural history, and this may seem

mysterious, even if not incredible.

Relatively little is said by Foot about moral epistemology. As I noted in Section 2.3,

Foot is predominantly read as a derivationist, and so it is generally assumed that her

moral epistemology is something along these lines: even non-philosophers are to

develop a (perhaps implicit) theoretical account of human nature and then to derive

conclusions from it about what they ought to do. But I have argued in the previous

chapter that Foot is not a derivationist.

Foot does speak briefly of moral education. She writes:

Human life, unlike the life of animals, is lived according to norms that

are known and taken as patterns by those whose norms they are. So we

have to teach children what they may and may not do. Nor could these

norms be taught simply by telling children that they are to be

courageous and ‘authentic’, however important it is to encourage them

to be daring and also to allow them to discover their true desires. The

norms to be followed must largely be formulated in terms of the

prohibition of actions such as murder and theft. (NG, 114)

The idea is nothing baroque. We do not give children discourses on human nature.

Moral education, rather, consists largely in telling children what to do and what not to

do. Maturity, perhaps, consists in children’s learning to weigh considerations against

each other and in their becoming able to determine what to do when confronted with

complex situations where their inclinations pull in different directions (cf. Section 4.6).

How might such a view be consistent with the idea that the virtuous life is natural to

human beings—as Foot attempts to argue in NG? Does not Foot depend here on the

assumption that children will be taught the right norms? How is that a matter of nature?

Well, the assumption is not arbitrary, if we agree with Anscombe that “it belongs to the

natural history of man that he has a moral environment” (2008, 224). Anscombe has in

mind the fact that human beings are social animals, which means that it is no accident

—it is natural—for human children to acquire a certain range of moral concepts in

terms of which moral norms are framed, as well as inclinations to respect those norms.

She is interested, in particular, in the way in which moral action-descriptions are “high-

level interpretations” of human behavior. It takes a certain sophistication to be able to

identify, e.g., someone’s picking some item up as stealing. And yet it is as ordinary as

anything that a child will acquire this sophistication and come to wield the concept of

theft competently—as ordinary as the fact that a child acquires the language of those in

its company. As Anscombe says:

Quite generally: to grow up as a child of normal intelligence in a human

society is eo ipso to be equipped with a range of concepts which form

the raw material for moral action descriptions, and in many cases to

acquire these as well, at least in a rough inchoate form. (2008, 225)

In the case of human beings, the division is false between what comes of nature and

what of family and society, for it is part of human nature to be part of a family and a

society and to react in certain ways to moral training. We may maintain, then, that it is

a part of human life to acquire the habits which are the virtues, even if this substantially

requires a training at the hands of one’s elders, and even if moral training is in many

cases poorly accomplished.

It is part of the human form of life to operate with, e.g., the concepts which

characterize justice. Thompson claims that the possibility of such a form of life “can

only be realized at all if its bearers grasp it other than empirically” (2022, 77). That

human beings acquire the relevant moral action descriptions as a matter of course in a

human upbringing is part of what constitutes human life as characterized by such ways

of going on as: doing what justice, courage, temperance, and prudence require. The

application of such action descriptions for which such an upbringing prepares one is

itself a manifestation of a distinctive kind of life, the human one.

We can, after the fact, recognize the necessity of such patterns of response for

the achievement of various goods—and that may tell us something important about the

virtues and their role in human life (cf. Müller 2018). As we saw in the previous

chapter, though, such considerations are not those that characteristically motivate the

virtuous agent. I have argued in this chapter, against Foot, that such considerations are

not even what secures the goodness of acting from the virtues. They could not be.

This point underwrites another sense in which human good is sui generis. Virtuous

patterns of motivation are constituted as part of human nature, as being naturally good,

in part by being naturally recognized as such by the virtuous individual. Compatibly

with neo-Aristotelian naturalism, they are not justified from without the agent’s

perspective.

Chapter 4: The Form of Practical Thought

Hegel seems to me to be always wanting to say that things which look

different are really the same. Whereas my interest is in showing that

things which look the same are really different. (Wittgenstein,

conversation with Maurice Drury, quoted in

Malcolm 1981)

4.1 Introduction

A virtue is a complex disposition. It is, inter alia, a disposition to respond to certain

sorts of situational reasons correctly and efficiently. The qualification is necessary here

because, as I noted in Section 2.7, a virtue is also a disposition not to respond to—even

not to notice—other sorts of features of one’s situation.

All the same, the sorts of reasons for which one acts are a central component of

virtue and thereby of practical rationality, or wisdom. The nature of reasons for action

is, of course, much debated (cf. Schroeder 2007; Alvarez 2010; Scanlon 2014). It is

helpful to frame discussion in terms of the considerations which collectively speak in

favor of some way of acting on a particular occasion—for indeed such considerations

are usually several, and not all of them are, individually, reasons for performing the

action in question. (Some such considerations might be, e.g., meansend beliefs or even

countervailing considerations.) That is to say, it is helpful to frame discussion in terms

of an agent’s rationale for an action, or in terms of her practical inference. (For stylistic

and dialectical reasons, I will mainly speak here of inferences rather than rationales,

but the meaning is meant to be the same. Cf. Section 1.4.) In this chapter, I consider the

structure of such inferences. What does it take for a practical inference to be good—to

be, in suitably broad senses, valid or sound?

This topic, the structure of practical justification, is enormous. I do not aim here

to treat all of its aspects. My intent is to survey those of its dimensions which are

especially relevant, in conjunction with what I have argued in previous chapters, to

sketching a picture of the wise individual. I accordingly set aside such topics as the

ontology of reasons and the nature of the conclusion of practical reasoning (though, as

I noted in Section 1.6, I will assume, for the sake of simplicity, that the conclusion of

practical reasoning is an action).

I will divide practical inference into two components, which I call instrumental

and situational inference. Instrumental inference, in which one discovers and takes

means to one’s ends, is the least controversial component of practical rationality,

although its nature is a subject of much dispute. I will be concerned to reject two

attempts to assimilate instrumental to theoretical validity, one (which I consider in

Section 4.2) focusing on inference to necessary means, the other (in Section 4.3)

focusing on inference to sufficient means. I argue that the notion of a sufficient means

is fundamental to the understanding of instrumental rationality and that there are

important differences between instrumental and theoretical validity.

Now, on the face of it, if what I have argued in Chapter 2 is correct—if there are

situational reasons—then not all practical inference is instrumental inference. In

Section 4.4, I consider an argument to the contrary derived from G. E. M. Anscombe.

Although I do not find that argument compelling, I will suggest that it raises the

important question of the role of desire in practical inference. In Section 4.5 and 4.6, I

will agree with Anscombe that it is neither false nor trivial to hold that, in a certain

sense, desire prompts action in every case—that desire is a sine qua non of action. This

fact helps us to distinguish between forms of practical failure, the cognitive and the

conative, which distinction amounts to the traditional distinction between practical

reason and the will.

4.2 Inference to Necessary Means

Discussions of instrumental rationality commonly begin with consideration of the

requirement that one take means that are necessary to one’s ends. Kant is a venerable

example in this regard; he wrote: “Whoever wills the end also wills (in so far as reason

has decisive influence on his actions) the indispensably necessary means to it that is in

his control” (2012, 4:417).

A more recent example is John Broome, who says that it is the “most

fundamental requirement of practical rationality” to “intend what you believe to be a

means implied by an end you intend” (2013, 159-160). A means implied by an end is a

condition without which the end would not obtain, the intending of which the agent,

moreover, takes to be that through which the end will be obtained, if indeed it is

obtained. More roughly but more intuitively, a means implied by an end is a means you

must take, and believe you must take, if you are to obtain your end. Why think that this

is the most fundamental requirement of practical rationality? On the face of it, at least,

the requirement lacks in generality of application. In many instances, it seems, we do

not think that any specific means is implied by our ends. One wants breakfast, for

example, and one might have it by having either eggs or yogurt or oatmeal.

Broome agrees. No specific means is implied by one’s end in such a case. Yet

Broome would insist that his fundamental requirement is still in play:

You will often be able to identify a number of specific alternative means

for achieving your end. That is to say, you believe that a disjunction m

or n or p or . . . constitutes an implied means to the end. Then rationality

requires you, if you intend the end and you believe the truth of the

disjunction is up to you, to intend the disjunction. (169)

A disjunction of specific means is implied by the end of having breakfast: the

disjunction of each thing one might eat. One then is required by rationality, on

Broome’s account, to intend the disjunction.

Intending the disjunction, of course, does not take one to intending any specific

means in particular, and thus it does not take one to action. This gap would be closed if

one among the disjuncts could be singled out. On Broome’s account, one of the

disjuncts is to be singled out by way of another requirement of rationality, according to

which you are “required to choose what you believe is the best of the alternative

means” (170). In this way, Broome intends to handle cases where no specific means is

itself necessary for one’s end.

What if some of the means are equally suitable? Broome does not say. But it

seems clear that then there could not then be a requirement to choose one of them in

particular but only a requirement to choose among them. Rationality would not require

any specific choice among equally suitable means. Sarah Paul makes this point

explicitly (after transposing it from the language of rational requirements to that of

reasoning):

In general, the performing of particular actions bottoms out in large part

with mere plumping for indifferent, adequate ways of getting it done.

And I suggest that once we reach the point of plumping for one of

whatever particular means are available

and perceived as equally acceptable, this should no longer be understood as an

exercise of reasoning, for it requires no further judgment of choiceworthiness.

(2013, 296).

Broome, unlike Paul, is writing explicitly of rational requirements rather than patterns

of inference. Rational requirements have some bearing on reasoning, though, since

Broome takes reasoning to be one of our ways of coming to satisfy rational

requirements. So although the correspondence between reasoning and rational

requirements may not be in every case neat and direct, the account suggests as

fundamental a certain picture: We begin with ends and (to the extent that we are

rational) infer what we must do as a means to obtaining those ends. In cases where it is

natural to say that we have multiple options, what is recognized is that taking one of

them (either m or n or p or …) is necessary for one’s end. If there is no uniquely best

option, then the selection of means is not inference; it is mere ‘plumping’.

Anscombe speculates about the attraction of such a picture and the prominence it gives

to the notion of a necessary means. The truth lying behind it seems to be: “If there is

practical inference, there must be such a thing as its validity. Validity is associated with

necessity” (2006a, 110). Those means that are necessary for the attainment of one’s

ends—those which one will have taken, if one achieves one’s end—are accordingly a

natural place to look, in thinking about the validity of practical thought. The impulse

seems to be to assimilate practical reasoning to ordinary deductive reasoning as much

as is possible. Hence Broome’s characterization of necessary means as means “implied

by” an end.

Is this a satisfactory account of the thinking of someone who takes some specific

means to an end? I want to argue that it is not. For the account lacks generality of

application even after Broome’s attempt to extend it to the case where various specific

means individually suffice for one’s end. For even in such a case, one need not (and

generally will not) believe that all of those means suffice for one’s end. One often will

not even be not aware of them all. Thus Broome’s disjunction generally is not a means

implied by the end, in Broome’s sense. Someone who seeks to make breakfast may,

simply, make some eggs, because making eggs is a means to making breakfast and it is,

say, the first thing one thinks to do. There need then be nothing wrong with the

inference to making (or intending to make) some eggs, even if it does not pass through

the recognition that making eggs or having oatmeal or having yogurt is necessary if

one is to have breakfast. If Broome’s requirement is fundamental, then there would

seem to be something, if not wrong, then peripheral about the case in which one simply

chooses a means that is sufficient but not itself necessary for one’s end. But such a case

is not peripheral at all. It is as ordinary as anything.

One might try representing the disjunction of specific means differently, in cases

where the agent (reasonably, it seems) does not consider every alternative. One could

say that what is a means implied by the end e is, say: p or some other means to e. But if

the move from the disjunction to intending that p is not in general to be reckoned

reasoning—as it seems it will not be, on Broome’s account, in the absence of the

judgment that p is the uniquely best means to e—then this suggestion would represent

the formation of an intention to bring about p or some other means to e as the

fundamental work that reasoning does in the selection of a sufficient means. But that

step is all but trivial. On the face of it,

I’d like to make breakfast.

Making eggs is a means to making breakfast.

So I’ll make eggs.

is a representation of a non-trivial bit of thought in which one engages in deciding to

make some eggs. (Some of the further questions which such a formulation raises will

be addressed below.) Broome’s prioritization of the notion of a necessary means

renders this inference questionable, elliptical, or non-central, while it seems to be none

of those things.

Although (as I have argued in Chapter 1) the course of an agent’s active reasoning is at

best a defeasible indicator of the content of the agent’s rationale, it is instructive in this

case to consider the fact that one often—even generally—acts without considering the

full variety of alternative means available. This suggests that the necessity of Broome’s

disjunction is not part of the justification of—the rationale for or practical inference to

—one’s action. (Of course, it would not often be mentioned if one were asked why one

took some specific means.) Broome and Paul both seem to privilege in their accounts a

certain kind of choice situation—one in which an agent has been apprised of all of the

options available to her before acting. In such a case, it looks like the only reasoning

for one to do is to select the best of those means, or one of the best. But common as the

assumption is that deliberation is a matter of choosing the best among several options,

the case in which one does so really seems to be special.

This makes sense if we consider what instrumental reasoning is. It is reasoning

aimed at the realization of some end. Fundamentally, in instrumental reasoning one

seeks to answer, for some action φ-ing, the question how to φ. One may come up with

several answers to this question and then choose among them according to some

principle. But one need not do so; one’s aim of φ-

ing, after all, will be served even if one does not apprise oneself of several candidate

means and choose among them. The work that is essential to deliberation, it seems, is

the discovery of means that suffice for, or at least contribute to, one’s end. (Several

sufficient means must be found, of course, if one is to compare them in the first place!)

What would lead one to suspect that inference to sufficient means should be

understood in terms of inference to necessary means? There is at least this connection:

one will consider what suffices for one’s end only if one needs to do so—that is, only if

one is unable to attain one’s end immediately, without deliberation. There is often, that

is, a necessity of finding some means to one’s end, if one is to obtain it at all;

deliberation, in other words, is frequently itself a necessary means, even where one has

a choice between what would more ordinarily thought of as one’s means. But it is not

necessary that one recognize and choose among all possible means to one’s end.

Now, if the fundamental form of instrumental inference is inference to

sufficient means, and not to necessary means, will we be able to understand inference

to necessary means? It seems that we will be able to do so. The felt pressure to intend

to ψ, when one wishes to φ and takes it to be the case that one will not φ unless one ψ-

s, is a result of the fact that any answer to the question of how to φ will include ψ-ing.

Some means are necessary but not sufficient. For instance, it is necessary, if one

is to save one’s baseball card collection, to enter the flaming house. Yet entering the

flaming house is not

enough. To save the baseball cards, one must also do some other things too, and it may

be doubtful whether anything within one’s power will, in conjunction entering the

flaming house, suffice for the saving of the baseball cards. They just might be doomed.

If it were evident that they were doomed, that nothing one could do would suffice for

one’s end, then it would not be reasonable to intend what was merely necessary for

one’s end. We see then that the notion of sufficiency must be in the background even

where there is inference to necessary means; one must at least have some plan, or think

or hope one can find some plan, for attaining one’s end, if one is to reasonably decide

to do even what is necessary for attaining that end.

4.3 Inference to Sufficient Means

How, then, should we understand inference to sufficient means? More particularly, how

should we understand the form that such an inference takes? G. E. M. Anscombe, in her

wide-ranging later paper “Practical Inference,” makes a proposal. She suggests that the

form is in fact just that of ordinary deductive inference—although the assimilation

works differently, of course, than what we have just seen attempted in regard to

necessary means. In this section, I will consider what leads Anscombe to this account

and ask whether it should be accepted.

In “Practical Inference,” Anscombe officially revises a view she defended in

Intention, concerning the nature of practical inference and its distinction from

theoretical inference. In Intention, Anscombe denies that a difference in content is

enough to distinguish practical from theoretical inferences. Instead, she writes:

There is a difference of form between reasoning leading to action and

reasoning for the truth of a conclusion. Aristotle however liked to stress

the similarity between the kinds of reasoning, saying that what

‘happens’ is the same in both. There are indeed three types of case.

There is the theoretical syllogism and also the idle practical syllogism

which is just a classroom example. In both of these the conclusion is

‘said’ by the mind which infers it. And there is the practical syllogism

proper. Here the conclusion is an action whose point is shewn by the

premises, which are now, so to speak, on active service. (2000, 60)

What does it mean to call the difference between practical and theoretical inference a

difference in form? The notion of form at work is logical; theoretical and practical

inferences differ in validity. They differ in what condition they need to meet to have the

kind of inferential goodness which goes by that title. On the assumption of the

premises, a theoretical inference needs to show the truth of a conclusion, while a

practical syllogism needs to show the point of an action. They differ also in what sort

of thing is their conclusion: on the one hand a belief, on the other an action.

So what Anscombe comes to relinquish is this idea that their respective sorts of

validity is what distinguishes practical from theoretical inferences. Why? Negatively,

Anscombe examines and finds wanting the attempts of philosophers such as Hare

(1964, 17-31) and Kenny (1966) to provide logics of practical inference (2006, 120-

123). Practical inference (which Anscombe here restricts to instrumental inference)

simply does not seem to bear a specifiable form of validity— and that seems odd, given

how straightforward it is (relatively speaking) to specify a form for deductive

inference.

Positively, Anscombe develops a particular example and professes to discover

in it that the form of practical inference is simply ordinary deductive validity. Her

example is of a man trying to assemble one of several committees (2006a, 123-127).

Her idea seems to be that, in such instrumental reasoning, what the man is looking to

do is to make it true that his end obtains—and he evidently will do that by satisfying

some sufficient set of truth conditions for the end. Accordingly it looks like what it is

for his practical inference to be valid is simply what it is for a corresponding theoretical

inference to be valid.

And so, Anscombe concludes, practical and theoretical inferences need not be

taken to differ in form. She writes that “[w]e can represent any inference by setting

forth a set of hypothetical considerations,” such as

If p, q.

If q, r.

so that the considerations comprising practical and theoretical inferences are “exactly”

or “just” the same and only differ in the “use” or “service” to which they are put. These

considerations, she says, could be used to assert r, if one can assert p, or to make r true,

if one can make p true (Anscombe 2006a, 128). In this sense, she comes to say,

practical and theoretical inference do not differ in form.

Anscombe has introduced two ideas here. Let us call the idea that practical and

theoretical inference do not differ in form formal monism. Let us call the idea that

practical and theoretical

inference are distinguished by the uses to which they are put—or by their teleology—

the teleological conception of practical inference.

Anscombe’s adoption of formal monism raises the question how much of Intention’s

conception of practical inference she means to give up. Indeed, Intention claims not

only that practical and theoretical inferences differ with respect to their form or

validity, but that the representation of action in practical thinking, which corresponds to

the premises of practical inferences, has a distinctive form. As Anscombe puts it,

“‘intentional’ refers to a form of description of events” (2000, 84; cf Thompson 2008,

85-146; 2011). It is hard to see how this claim could be independent of the thesis that

practical and theoretical inferences differ in form. For, according to one influential and

plausible way of thinking about the matter, the logical form of a proposition just is its

structure insofar as it bears on inference, and it is only in light of propositional

structure that one can frame the question of how an inference’s constituent propositions

need be related if it is to be valid (cp. Frege 1997, 48-49).

The claim that ‘intentional’ refers to a form of description of events plays a key role in

Intention. At what is evidently its climax, Anscombe claims that where (a) the

description of an event is of a type to be formally the description of an

executed intention (b) the event is actually the execution of an intention

(by our criteria) then the account given by Aquinas of the nature of

practical knowledge holds: Practical knowledge is ‘the cause of what it

understands’, unlike ‘speculative’ knowledge, which ‘is derived from

the objects known’. This means

not just with respect to is object and its end, but also with respect to its “mode”: knowledge is practical

with respect to its mode when it concerns “how” an end is to be brought about.

more than that practical knowledge is observed to be a necessary

condition of the production of various results; or that an idea of doing

such-and-such in such-andsuch ways is such a condition. It means that

without it what happens does not come under the description—

execution of intentions—whose characteristics we have been

investigating. (2000, 87-88)

When someone exercises practical knowledge, “what he effects is formally

characterised as subject to our question ‘Why?’ whose application displays the A—D

order which we discovered,” and the man thinks of his action as such (2000, 88).

According to Intention, the agent’s thought about her action itself has a distinctive

form, which in part means that it stands in distinctive inferential relations. So this idea

seems, at least, inseparable from a formal pluralism about theoretical and practical

inference. Yet Anscombe, while signaling a change in view, does not discuss the extent

to which the core claims of Intention may be retained.

Will Small has argued in some detail that Anscombe’s later account of inference

in fact does give too much up. Core to his argument is the need for distinguishing, as

Anscombe in her later account does not, between p’s being a sufficient means to q and

p’s being sufficient for q (2021, 266-267). For instance, my running quickly is sufficient

for my running, but is not of course a means to my running; and in general there can be

conditions (say, that p) that will obtain only if q (so that it is true that if p, then q) that

are not means to q. It is not, Small concludes, the material conditional ‘if p, then q’ but

the practical conditional “doing A is a means of doing B” that must figure as premise

in practical inferences, and “the concept of a ‘way’ or ‘means’ is a formal concept of

practical thinking: a practical conditional ‘connective’ that is a specific determination

of ‘ ’” (2021, 268).

We are to understand the meaning of this means-end premise “doing A is a way of

doing

B” precisely by considering the aim of practical thinking: action. Small writes:

The teleological conception of practicality takes seriously the idea that

practical reasoning is for the sake of realizing in action the objective

that is its starting point. And taking this seriously requires more than

just asserting that the conclusion ‘is an action’: we need to think about

what intentional action is, and what its subjective conditions of

possibility are, in order to understand what kind of practical thought it is

the function of practical reasoning to deliver. (2021, 272)

He sees this task better executed by Intention than by “Practical Inference.” On his

account, such a premise as “doing A is a way of doing B” is formally distinct in a

variety of ways from a mere material conditional. For instance, it is only true, and only

practical, if doing A is itself practicable for the agent; it is only true if it is believed by

the agent; and its truth must be certified by perception (Small 2021, 267-273).

In Intention, Anscombe argues that practical inference, though it has a

characteristic content (namely, it concerns future contingents within one’s power), is

not distinguished by its content but rather by its form. It may be, likewise, that practical

inference has a characteristic form, even though it is distinguished not by its form but

by its teleology. The distinction in use between practical and theoretical inference may

indeed explain why practical inference has a characteristic form:

Practical reasoning itself distinguishes itself from theoretical reasoning

through its teleology. Thus far Anscombe is correct. But that practical

reasoning distinguishes itself from theoretical reasoning by its teleology

does not mean that it does not have a distinctively practical form, or that

it is a matter of indifference whether this is acknowledged. It has a

significantly different form: and its formal differences are determined

by its distinctively practical function. (Small 2021, 283)

An analogy may be helpful in this regard. A stone and a hammer may both be used to

hammer. But the hammer, unlike the stone, is structured the way it is because it is for

the sake of hammering nails. The stone and the hammer, one might say, differ in in the

character of their teleology; they differ as regards whether their being for the sake of

hammering is accidental or intrinsic to them. A stone is used, by some agent, to

hammer, while the hammer is such as to be suitable for hammering. The same

distinction might be employed in the context of thought and talk. I might use a word

because I think that it will evoke a certain thought in you in particular, but that is very

different from using a word to mean what the word standardly means. Anscombe, in

her later paper, effectively gives practical thought an accidental teleology, for she

assumes that one and the same inference might be put to practical or theoretical use.

But if Small is right, then we might say that the teleology of practical thought is in fact

intrinsic to its form, and so that the teleological conception of practical inference is not

only separable from formal monism, but properly understood serves to show why

formal monism is false.

Thus, while the notion of a sufficient means is central to instrumental inference,

it cannot simply be understood simply in terms of logical sufficiency. The form of

instrumental inference must rather be put this way:

I’ll φ. ψ-ing is a

means of φ-ing.

So I’ll ψ.

4.4 Anscombe on Backwards-Looking Motives

The formal concept of a means characterizes at least some of the premises of practical

inferences. What other premises might practical inferences might have? Situational

reasons are obvious candidates. Those are, again, considerations of particular facts

about one’s situation such as that one promised to φ, that A is in need, that X is

dangerous, and so on (cf. Section 2.4).

Anscombe considers and rejects the idea that there is inference from situational

reasons— or what she called backward-looking motives. On her view, in both Intention

and the later “Practical Inference,” practical inference is calculative or instrumental.

She indeed acknowledges that some of what seems to be practical inference does not fit

the schema of instrumental reasoning: ‘I want a Jersey cow; there are good ones in the

Hereford market, so I’ll go there’ would seem to be practical reasoning too. Or ‘If I

invite both X and Y, there’ll be a strained atmosphere in view of what X has recently

said about Y and how Y feels about it—so I’ll just ask X’. Or again ‘So-and-so was

very pleasant last time we met, so I’ll pay him a visit’. (2000, 62)

Yet her ultimate stance is that these are not inferences, for they do not involve

“calculation,” and “[t]he mark of practical reasoning is that the thing wanted is at a

distance from the immediate action, and the immediate action is calculated as the way

of getting or doing or securing the thing wanted” (2000, 65-67, 79).

Anscombe speaks of “the thing wanted.” She indeed argues that desire is necessary for

action. In regard to the above examples, she writes:

Now Aristotle would have remarked that it is mere ‘desire’ in a special sense

(ἐπιθυμία) that prompts the action in the last case; the mark of this is that the

premise refers to something merely as pleasant. The point that he is

making here is, however, rather alien to us, since we do not make much

distinction between one sort of desire and another, and we should say:

isn’t it desire in some sense [ὄρεξις]—

i.e. wanting—that prompts the action in all the cases? And ‘all cases’,

of course, includes ones that have as large an apparatus as one pleases

of generalisations about morals, or medicine, or cookery, or methods of

study, or methods of getting votes or securing law and order, together

with the identification of cases. (2000, 62-63, emphasis added)

Further, she writes:

I am not saying that there cannot be any such thing as moral general

premises, such as ‘People have a duty of paying their employees

promptly’, or Huckleberry Finn’s conviction, which he failed to make

his premise: ‘White boys ought to give runaway slaves up’; obviously

there can, but it is clear that such general premises will only occur as

premises of practical reasoning in people who want to do their duty.

(2000, 78)

This claim is controversial. For while Anscombe clearly means to be saying something

substantive here, it has been argued by a variety of philosophers that it is either false or

trivial to hold that all action is prompted by desire.

I will return in what follows to the question of what role desire has in practical

inference and in action. For now, I want to note that despite Anscombe’s holding that

desire is necessary for practical inference, she denies that any given practical inference

must have a premise stating that one desires something. “[I]t is misleading,” she says,

“to put ‘I want’ into a premise if we are giving a formal account of practical

reasoning.”

To understand this, we need to realise that not everything that I have

described as coming in the range of ‘reasons for acting’ can have a

place as a premise in a practical syllogism. E.g. ‘He killed my father, so

I shall kill him’ is not a form of reasoning at all; nor is ‘I admire him so

much, I shall sign the petition he is sponsoring’. The difference is that

there is no calculation in these. The conjunction

‘so’ is not necessarily a mark of calculation. (2000, 65)

The argument that Anscombe is making here might seem to beg important questions. Is

“calculation” simply defined in terms of instrumental inference? If so, then whence the

contestable requirement that practical inferences involve calculation? And if not, then

who is to say that inference from situational or backwards-looking reasons is not

calculation?

Anscombe’s argument does seem to me to be question-begging in this way. She

says too little for it to be compelling. But I think it is possible to develop her remarks

into another sort of argument, if only a negative one, against counting situational

reasons as premises to practical inferences.

Anscombe makes the point that both “I want a Jersey cow” and “He killed my

father” can be reasons for action, in the sense that they are answers which do not refuse

the question ‘Why?’ by which she delimits the class of intentional actions. (I will

shortly have more to say about Anscombe’s question ‘Why?’.) Nevertheless it would

be a mistake to include a fact about what one wants as a premise in a practical

inference. That is what Anscombe is here trying to show; but if we established that

point independently, then we might appeal to it to generate some grounds for

questioning in addition whether situational reasons are premises to practical inferences.

That is the argument I will now develop, on Anscombe’s behalf.

How could we establish independently that it is a mistake, “if we are giving a

formal account of practical reasoning,” to include a fact about what one wants as a

premise in a practical inference? Well, we can consider a case where a fact about what

one wants is a premise to a practical inference, and see that it is not what we were after,

in accounting for the necessary role of desire in action. Consider this example that

Anscombe in “Practical Inference” attributes to

Anselm Müller:

Anyone who wants to kill his parents will be helped to get rid of this trouble by

consulting a psychiatrist.

I want to kill my parents.

If I consult a psychiatrist I shall be helped to get rid of this trouble.

NN is a psychiatrist.

So I’ll consult NN. (2006a, 115)

Here, the agent’s wanting something is a premise to a practical inference, and yet the

tendency of the inference is not to the obtaining of the thing said to be wanted, but to

the avoidance of it. The fact that one wants to kill one’s parents is part of the fact

situation, as it were. In other words, “the thing wanted,” in the relevant sense, is not: to

kill one’s parents. It is, rather: to get rid of this trouble. The example shows that the

role of the thing wanted is not aptly captured by including the fact of its being wanted

as a premise. So—according to the argument I am imagining—it is a mistake to hold

that if something is a reason for action (in the sense of being an answer that does not

reject Anscombe’s question ‘Why?’) then it is a premise to a practical inference.

So conceived, the argument is purely negative. Its conclusion could be admitted while

still holding, on different grounds, that there are inferences from situational reasons. I

would like to produce such grounds. Namely, I want to argue that Anscombe herself is

committed to holding that situational (or backwards-looking) reasons are premises of

arguments, for otherwise she will lack the resources to identify them as a positive

answer to her question ‘Why?’.

Anscombe, famously, seeks to characterize the class of intentional actions as those to

which a certain sense of the question ‘Why?’ is applicable (2000, 9-28). She specifies

the sense of this question by surveying the sorts of positive answers it can receive. Its

positive answers, intuitively, give reasons for action, although Anscombe takes that

characterization to be unhelpful for her project—for explaining what a reason for

action is and explaining her sense of the question ‘Why?’ are commensurate tasks (cf.

2000, 10).

One type of positive answer to the question ‘Why?’ is, as it were, future-tensed. To

recognize such an answer as an admission of the question ‘Why?’, Anscombe thinks it

suffices to note its future tense—its reference to an unrealized future possibility

(relative to the time of the action). This on its own distinguishes an answer identifying

what she calls a forward-looking motive (an intention or aim) from other answers to the

question ‘Why?’. But the same is not true of the past tense and what Anscombe calls

backwards-looking motives. Most importantly, a backwards-looking motive must be

distinguished from what Anscombe calls a mental cause—for example:

—Why did you knock the cup off the table?

—I saw such-and-such and it made me jump. (2000, 16)

So the class of backwards-looking motives is not sufficiently distinguished by appeal to

the past tense of the answer to the question “Why?”. Nor does so-called knowledge

without observation suffice as a differentia, for in the cases of both backwards-looking

motivation and of mental causes,

the agent “knows without observation” ‘why’ she acted as she did—she is able to say,

straight off, without needing to check, why she killed him or knocked the cup off the

table.

What does distinguish backwards-looking motives from mental causes, Anscombe

says, is the involvement in the former of the ideas of good and evil (2000, 22). We can

understand what she has in mind if we attend to her examples of backward-looking

motives: revenge, gratitude, remorse, and pity. We have already imagined a bit of

dialogue characteristic of revenge, and it is easy to imagine dialogue corresponding to

the rest of Anscombe’s backwards-looking motives. In each case, the intelligibility of

the answers to the question “Why?” depends on certain of the agent’s beliefs about

good and evil. Revenge, for instance, requires that you take yourself to have been

harmed by someone in some way; if you come to be persuaded that you were not after

all harmed, then the motive ceases to make your action intelligible. And it requires that

you take what you try to do in response to be harmful to your target; if you come to be

persuaded that your action will not harm your target, then again your answer ceases to

rationalize your action. Mental causality is not like this. One’s account of why one

jumped (“I saw a face in the window”) is not undermined if someone says that there

was no face in the window, or that such face in the window is no occasion for jumping.

Anscombe does not say whether she thinks her list of backwards-looking motives is

exhaustive. It seems not to be. Indeed the motives she has selected seem to form only a

very small sample of what should be regarded as backwards-looking motives. What I

have called situational reasons seem to be examples of backwards-looking motives. As

we have seen, Foot’s work is itself

based on Anscombe’s (1969) treatment of promising. To fulfill a promise is to act for a

certain kind of reason, in conformity with a certain language-game, whose structure

basically takes this form:

—Why are you φ-ing?

—Because I promised A that I would φ.

Obviously, this answer does not cite a mental cause. But good and evil are not involved

here—not, at least, in the straightforward and clear way in which they are involved in

revenge, gratitude, remorse, and pity. To take another example from Intention, one

might do something “because he told me to” (2000, 23). Anscombe wonders whether

this is a cause or a reason, and she concludes that it depends on the case. If so, then

when it is a reason, what kind of reason is it? It must be a backward-looking reason,

but again good and evil are not straightforwardly involved in it. So the involvement of

the ideas of good and evil are not necessary for backwards-looking motivation. Nor, for

that matter, is it sufficient. The thought of all the evil in the world, for instance, or of

the horrible thing A did, could very well serve as mental causes in Anscombe’s sense.

Yet the distinction between backwards-looking motive and mental cause is evidently a

real one. What does distinguish backwards-looking motives from mental causes, I want

to suggest, is that backwards-looking motives can sensibly be argued against. In

Anscombe’s examples, the way in which they would be argued against is by calling

into question certain facts about good or evil. In other cases, they would be undermined

in different ways, depending on the pattern of backwardlooking motivation in question.

For instance, the following might undermine someone’s doing something because she

promised:

—A released you from your promise—didn’t you hear?

—Surely A wouldn’t mind, given the circumstances, if you don’t keep the promise.

—That hardly counts as keeping your promise.

The specific examples Anscombe considers—again, revenge, gratitude, remorse, and

pity—make sense only if the agent is taken to have certain thoughts about the good and

evil of actions. That is, good and evil are part of the form the bearing of which

constitutes motivation as revenge, gratitude, remorse, or pity. This is why contesting

those thoughts about good and evil can call into question the agent’s motive and can

potentially undermine it. But if that is right, then why should not other ideas be

characteristic of other patterns of motivation? Even Anscombe’s examples bear out,

then, that backward-looking motives are distinguished from mental causes in that they

can be argued against, on the basis of whether they really conform to some pattern

characteristic of them.

This is to say that it is a distinguishing mark of situational or backwards-

looking reasons that they are in the space of arguments, for it is arguments that can be

argued against. In other words, the backward-looking or situational inferences that

Anscombe said were not reasonings are reasonings—they are syllogisms, or rationales,

or arguments for actions. So Anscombe should acknowledge that it can be inference to

conclude that it is worthwhile to φ on the basis of facts rather than on the basis of the

conduciveness of φ-ing to one’s unrealized aims.

4.5 Situational Inference

Anscombe calls the motives or reasons in question ‘backwards-looking’ because she

takes them to be past-tense in expression, but as Müller (2011) points out, this seems

not to be essential to the relevant form of motivation. One might do something

because, e.g., A needs help. Accordingly, in

Chapter 2, I denominated such reasons situational. We can say, then, that a situational

inference is an inference of the abstract form:

So I’ll φ.

It is an inference from some fact, a situational reason, to an action (or, if you like, to an

intention, or to an evaluative belief; cf. Section 1.6).

As I indicated at the outset, situational and instrumental inferences are components of

practical inference. They are generally, that is to say, combined: p. So I’ll φ. ψ-ing is a

means of φ-ing.

So I’ll ψ.

There are various more determinate forms of situational inference; they may be

enumerated by considering the sorts of facts on the basis of which one must act, if one

is to count as acting from a motive such as justice, fairness, charity, revenge, gratitude,

and so on. To act from such motives requires treating one kind of reason or another as

ultimate. One can give someone what she is owed for reasons of convenience, say, but

one only counts as doing it from justice if one’s reason is that she is owed it.

Is situational inference a distinctive kind of inference indeed, or might it be

assimilated to instrumental inference? Take, for instance, this ostensibly situational

inference:

I promised to φ.

So I’ll φ.

This might be cast in non-situational form by characterizing the keeping of one’s

promise as one’s end:

I’d like to keep my promise to φ.

φ-ing is a means to keeping my

promise to φ.

So I’ll φ.

A parallel treatment might be proposed for each type of purportedly situational

inference.

Müller resists such assimilations. He writes:

[T]he two conceptions of your reason for doing what you promised are

not on a par. The teleological [or instrumentalizing] interpretation of

promise-keeping identifies an aim of φ-ing—viz., keeping the promise

—that can itself be specified only in terms of an already acknowledged

backward-looking reason. In the specification of this backward-looking

reason, on the other hand, no comparable reference to a forward-looking

reason is involved. Even if the rationality of promise-keeping is viewed

as forward-looking, the priority of its backward-looking rationality has

to be acknowledged. (2011, 260-261)

There seem to be two ways of understanding this point. On the one hand, one might

think that the existence of a concept such as that of taking revenge—as the sort of thing

that might be made into one’s end—depends itself on the existence of a pattern of

inference according to which a person harms someone simply because he has harmed

her. On the other hand, one might think that the

ascription to an agent of an instrumentalizing, non-situational form inference itself

depends on the agent’s also, and more fundamentally, being motivated by a situational

reason.

Both ways of understanding the point seem to me to be correct. One can make

something like the taking of revenge one’s end. But this is only because there is such a

thing as taking revenge: to harm someone because he harmed you. Moreover, you will

only succeed in taking revenge if you are motivated by the relevant sort of situational

reason. One might, a bit grotesquely, imagine that someone wishes to take revenge

(perhaps she wants to know what it feels like), but simply is unable to possess the

requisite motivation; she can rehearse the above syllogism, but ultimately she kills the

man not because he killed her father (though he did and she knows it) but only out of a

fruitless attempt to take revenge. She does not, as it were, feel it. She will not really

count as taking revenge, due to the absence of the relevant situational reason.

That said, I consider to be relatively academic the question whether situational

inferences can be paraphrased without remainder in non-situational terms. What is

important for the understanding of practical rationality is that, however situational

inferences are to be understood, many of our ends are situation-sensitive. Even if the

paraphrases are acceptable, it is still true that there are cases where my end is, say, to φ,

because I recognize that p—I wish to kill him, for instance, because he killed my

father, or I wish to help her, because she is in need. This phenomenon is sufficiently

distinctive to merit its own name.

4.6 Faults in Judgment

The stock examples we have seen so far might suggest that situational inference is a

relatively straightforward matter. I promised to φ, so I’ll φ; she deserves X, so I’ll give

it to her; and so on.

There might seem to be little that can go wrong in situational inference—besides not

drawing the relevant inference at all, when one should, or drawing an inference that

one shouldn’t, like the inference that characterizes revenge. Perhaps that is why

Anscombe denies that backwards-looking motivation involves ‘calculation’. In any

case, this impression is mistaken. The drawing of a situational inference is sometimes

complex and difficult.

I will illustrate what I have in mind here by example. Suppose that you have

damaged someone else’s property. In recognition of this, you might decide to give

recompense. But what sort of recompense are you to give? Repairing the damage might

be impossible. Doing nothing will in many cases be inappropriate. But if you have not

altogether destroyed the property, then replacement may well not be appropriate either

(and anyway, replacement will sometimes be impossible). Sometimes an apology will

suffice, but not always. So what are you to do?

In the abstract, there is evidently no answer to give to this question. It depends

on the facts. And there will be no simple way of specifying just how it depends on the

facts. Yet in at least a wide range of cases, agents will agree that at least certain sorts of

responses are inappropriate. In other cases, there will be less consensus, but still space

for right and wrong; those who are capable of making such determinations we credit

with sound judgment or—I want to say—wisdom.

Let us put the point this way. There are relatively abstract forms of situational

inference.

For instance:

I damaged A’s property.

So I’ll give recompense.

Such a pattern of inference cannot be applied except by being made more specific—for

there is no such thing as simply giving recompense; there is only giving recompense in

a specific sort of way, in light of a specific kind of wrong one takes oneself to have

committed. So, for instance, one might infer:

I scratched the chassis of A’s bicycle.

So I’ll profusely apologize and inquire whether I can somehow make up for it.

Some attempts at instantiating the pattern will be inappropriate. For instance:

I scratched the chassis of A’s bicycle.

So I’ll replace the whole thing.

This is, presumably, as far as justice is concerned, excessive. The inference in question,

accordingly, might be thought to possess a kind of invalidity; the response does not

actually ‘follow’ from the relevant situational consideration, though it is meant to do

so, and to certain agents might seem to do so.22

There is another—related—dimension of complexity involved in this example and in

situational inference generally. As I’ve said, the appropriate response depends on the

facts. It is an oversimplification to cite a single situational reason in any given case, for

there are generally other relevant facts which will bear upon the appropriateness of

one’s response. Whether it is appropriate to give a certain kind of recompense depends,

for instance, on one’s ability to do so. And whether it is appropriate to give recompense

now as opposed to, say, later, depends on what else is going on in one’s situation.

Perhaps one’s resources and attention are, at the moment, more urgently required by

one’s child. In general, there may be rival situational considerations in play in a given

22 An equivalent point could be made if one favors an instrumentalizing paraphrase of situational

inferences, of the sort mooted in the previous section. We might then formulate the inference:

I’d like to give recompense for scratching the chassis of A’s bicycle.

Profusely apologizing and inquiring whether I can make up for it is giving recompense.

So I’ll profusely apologize and inquire whether I can make up for it.

Here we must say: what calls for sound judgment is the recognition of the truth of the relevant

instrumental belief. In either case, it may be far from straightforward to recognize what is the thing to

do, on the basis of what one’s situation is like.

situation which suggest altogether different sorts of response. It takes wisdom to be

able to see what to do when confronted with a complex situation.

Adapting a term from John McDowell, we might say that the canons of validity

of situational inference are not, in general, codifiable (2001a, 57-58). As he says:

If one attempted to reduce one’s conception of what virtue requires to a

set of rules, then, however subtle and thoughtful one was in drawing up

the code, cases would inevitably turn up in which a mechanical

application of the rules would strike one as wrong—and not necessarily

because one had changed one’s mind; rather, one’s mind on the matter

was not susceptible of capture in any universal formula. (2001a,

58)

But this lack of codifiability does not undermine the possibility that there are correct

and incorrect responses to one’s situation. We would face, no doubt, parallel difficulties

in theoretical contexts, if we tried to draw up guidelines for deciding what to believe—

but that does not defeat the prospect of soundness in theoretical judgment.

Let us summarize this discussion by saying that one kind of fault which

someone can commit in the drawing of a situational inference is a fault in judgment.

Someone commits a fault in judgment if she φ-s because p while φ-ing is not really an

appropriate response to the fact that p—if “p, so I’ll φ” is an invalid inference.

4.7 Faults in Motivation

As I noted at the beginning of the previous section, faults in judgment seem not to be

the only possible fault in situational inference. An agent can also be inadequately

motivated; she can, despite appreciating a situational reason and the availability of

some relevant inference, fail to draw it. The fact that this is a distinct sort of mistake,

and needs to be recognized as such, is, I want to say, part of the basis for drawing some

important distinctions made within the Aristotelian tradition: distinctions between

wisdom and virtues of character, between wisdom and cleverness, and between

practical reason and the will.

On what is called the neo-Humean theory of motivation, actions are caused by

reasons for action, where a reason for action includes a desire or some other “pro-

attitude” (Davidson 2001; Smith 1987). I am interested here not in the neo-Humean

theory but in the prominent reaction against it. Namely, it has been frequently

suggested that it is either false or trivial to hold that desire is necessary for action. T. M.

Scanlon, for instance, writes:

Rationality involves systematic connections between a person’s

judgments and his or her subsequent attitudes. A rational person who

judges there to be sufficient grounds for believing that P normally has

that belief, and this judgment is normally sufficient explanation for so

believing. There is no need to appeal to some further source of

motivation such as “wanting to believe.” Similarly, a rational person

who judges there to be compelling reason to do A normally forms the

intention to do A, and this judgment is sufficient explanation of that

intention and of the agent’s acting on it (since this action is part of what

such an intention involves). There is no need

to invoke an additional form of motivation beyond the judgment and the

reasons it recognizes, some further force to, as it were, get the limbs in motion. (1998,

33-34) The idea is that rationality, whether theoretical or practical, consists in the

obtaining of the right sort of connections among one’s attitudes. If one is rational, then

one believes what one takes one to have reason to believe, and one does (or at least

intends to do) what one takes one to have reason to do. There may of course be those

who do not so believe and so act; we call those people (in what is perhaps a technical

sense) irrational. But we do not need to appeal to some other factor, desire, to explain

why someone believes and does as she, rationally speaking, should. Rationality simply

is believing and acting as one, rationally speaking, should. This is especially clear,

Scanlon points out, in the theoretical case. Why think it is any different in the practical

case?

Of course, there does seem to be something right in saying, quite generally, that if A φ-

s, then she wanted so to φ. That is, someone’s φ-ing intentionally simply is a ground

for saying that she wanted to φ. It is such a ground even if A finds φ-ing distasteful or

in some other way unattractive—even if she φ-s with regret, so that in what is evidently

another sense of ‘want’, we would say that she did not want to φ.

Accordingly, there are at least two ways of understanding the claim that desire

is necessary for action, corresponding to two senses of ‘desire’ and ‘want’; and this

distinction in senses of

‘desire’ might be drawn in multiple ways. We saw Anscombe above distinguish what

the ancient Greeks called ἐπιθυμία (or appetite) from ὄρεξις (or desire in general).

Scanlon himself isolates a notion he labels “desire in the directed-attention sense.” One

desires to do something in the directed-attention sense not only if one is inclined

towards doing it but, for instance, if one constantly thinks about it, has trouble directing

one’s attention away from it, feels a so-called prick of desire at the thought of it, and so

on (1998, 39-41). And then Thomas Nagel, in a classic discussion, draws a distinction

between motivated and unmotivated desires—those which are themselves had, and

those which are not themselves had, for reasons. (1978, 29)

These distinctions are, of course, not quite equivalent, but they might be taken

to have a similar force: In each case, it would be substantive, though false, to say that

desire in the narrower of the senses is necessary for action. Here is how Nagel puts the

point:

The claim that a desire underlies every act is true only if desires are

taken to include motivated as well as unmotivated desires, and it is true

only in the sense that whatever may be the motivation for someone’s

intentional pursuit of a goal, it becomes in virtue of his pursuit ipso

facto appropriate to ascribe to him a desire for that goal. But if the

desire is a motivated one, the explanation of it will be the same as the

explanation of his pursuit, and it is by no means obvious that a desire

must enter into this further explanation. (1978, 29)

Similarly, what Aristotle calls appetite and what Scanlon calls desire in the directed-

attention sense seem not to be present wherever there is action. So it is false to say that

desire in any of the narrower senses is necessary for action.

On the other hand, though, it is true, but apparently trivial and uninteresting, to

say that desire in any of the contrastingly wider senses is necessary for action; it is

ascribed, as Nagel says, simply on the basis of the fact that the agent is pursuing a goal,

and does not obviously enter independently into the explanation of the agent’s pursuit.

From the fact that desire in the wider of the senses is ascribed wherever there is action,

though, it does not follow that it is otiose in the explanation of action. For if desire is a

sine qua non of action, then it may be that one explanation of why someone didn’t φ is

that she didn’t—in the wider sense—want to.

And it seems that we should recognize this kind of explanation. It is as recognizable as

the possibility of akrasia, or weakness of will. As we saw above, there is a possibility

of cognitive failure, fault in judgment, in the drawing of a situational inference—the

possibility of someone’s responding in some way, but wrongly, to a situational reason.

On the face of it, fault in judgment is not the only sort of failure one can commit;

sometimes, one might recognize that some situational consideration calls for response,

but fail to do anything in response to it. One is not, so to speak, willing so to act.

Let us say, then, that in addition to faults in judgment, there are faults in

motivation. Someone commits a fault in motivation if she thinks (perhaps wrongly) that

she ought to φ because p but does not. (The ‘ought’ here is the conclusive ‘ought’ of

deliberation; cf. Section 2.6.) So, to stick with our example from the previous section,

someone who makes inadequate recompense for damaging someone’s property

commits a fault in judgment, while someone who makes no attempt at recompense at

all may commit a fault in motivation. A fault in judgment is a kind of invalidity in one’s

situational inference. A fault in motivation is a failure to draw any relevant inference. It

is arguably a distinguishing mark of practical inference that faults of motivation are

possible.

This distinction sheds some light on Aristotle’s defense of the unity of the

virtues, the thesis that the virtues are mutually entailing, that having one means having

them all. His argument concerns the relationship between wisdom (an intellectual

virtue) and the virtues of character (such as courage, temperance, and justice). Aristotle

says that the former requires the latter, and the latter the former (2002, 1144b18-33). I

wish to formulate the argument in terms of what the different virtues rectify or make

right. Wisdom ensures that one does not commit faults in judgment, that one recognizes

the appropriate responses to the relevant facts. A virtue of character, by contrast, is

ascribed on the basis of one’s motivation to act in accordance with specific patterns of

situational inference. That is why, if virtues are characterized as dispositions to respond

correctly to a certain class of situational reasons, then the possession of any virtue of

character requires the possession of wisdom; one will not be responding correctly if

one responds inappropriately, for instance by giving someone less than what she is due.

The opposite entailment, from the virtue of wisdom to each virtue of character, can be

located in the fact that virtues are general dispositions and must, therefore, prepare an

agent to respond to potentially complex situations in which a variety of considerations

are at stake (cf. Müller 2004b). So wisdom and the virtues of character are mutually

entailing.

Nevertheless, even where one commits a fault in judgment by, say, giving

inadequate recompense to someone, one is at least attempting to give to her what is

due; there is sense to be made then in the ascription of something like virtue to

someone who is motivated to act in accordance with a certain pattern of situational

inference but is not good at finding the appropriate response. Such a person has, say,

justice, but in a defective sense.

This account of the distinction between wisdom and the virtues of character

also dovetails nicely with Aristotle’s distinction between wisdom and cleverness—the

capacity to find means to one’s ends (2002, 1144a25-26). Aristotle says that wisdom

depends on but is not the same as cleverness, for cleverness can exist in the vicious. It

can look as though, for Aristotle, wisdom simply is cleverness in a person with the

virtues—a person who has the right ends. The fact that the wisdom requires the virtues

of character then looks like mere stipulation. But on my view, what wisdom does, over

and above cleverness, is enable one to hit the mean in the ends one sets for oneself—it

enables one to recognize what, exactly, is an appropriate response to one’s situation.

This is a substantive task that cleverness does not accomplish. The wise and the unwise

in a sense do—but in another important sense do not—have the same ends; they may

both be motivated to act, say, justly, but what the unwise person tries to do will not

really be what justice requires in his situation.

4.8 Practical Reason and the Will

Now, McDowell parlays his denial of codifiability into, in effect, a rejection of the

distinction between wisdom and the virtues of character—for he defends the Socratic

thesis, against which Aristotle himself sets himself, that virtue of character is a kind of

knowledge. Here is McDowell’s case for doing so:

On each of the relevant occasions, the requirement imposed by the

situation, and detected by the agent’s sensitivity to such requirements,

must exhaust his reason for acting as he does. It would disqualify an

action from counting as a manifestation of kindness if its agent needed

some extraneous incentive to comply with the

requirement—say, the rewards of a good reputation. So the deliverances

of his sensitivity constitute, one by one, complete explanations of the

actions that manifest the virtue. Hence, since the sensitivity fully

accounts for its deliverances, the sensitivity fully accounts for the

actions. But the concept of the virtue is the concept of a state whose

possession accounts for the actions that manifest it. Since that

explanatory role is filled by the sensitivity, the sensitivity turns out to be

what the virtue is. (2001a, 52)

The concern here is similar to Scanlon and Nagel’s. It seems that an agent’s

appreciation of the reasons available to him can fully explain his response to them; in

the case of virtuous activity, McDowell adds, this even seems to be required, since

appeal to an extraneous factor would seem to disqualify the agent’s motivation from

being virtuous. Perhaps an extraneous factor may need to be cited where things go

wrong, but that should not be taken to shed light on what is necessary for the normal

exercise of the power of practical reason. So says Sebastian Rödl: In general, a power

of knowledge [in this case, practical knowledge] does not yield knowledge when its

exercise is impeded. And it may be that the will is impeded in its proper exercise in

cases in which one fails to do the thing from the willing of which one reasons

practically. Then the fact that it may happen that I reason practically from my will to do

A without doing A has no tendency to show that, when in fact I am doing A through

reasoning, my reasoning does not on its own account for this. (2011, 221)

Rödl represents a Kantian sort of opposition to distinguishing between practical reason

and the will—for it was Kant who proclaimed that “the will is nothing other than

practical reason” (2012,

4:412; cf. also Fix 2018).

There is a question, of course, of what we even mean when we speak of the

distinction, or indistinction, of powers like practical reason and the will. In speaking of

such matters, we are evidently in at least semi-technical territory. I take it that powers

are coordinate with certain sorts of activity, for a power is a power to do (in a broad

sense) something. Accordingly, the fact that faults in judgment and faults in motivation

are distinct sorts of fault seems to me to suffice for distinguishing powers in which to

locate these distinctive sorts of failure, which can come apart. Part of what motivates

the Kantian and neo-Kantian identification of practical reason with the will is, I think

it, the sense that the activities of these two powers do not come apart. When their

activities are identified as, say, practical thought and desire, they sound distinct; but the

case for identifying practical reason and the will seems stronger when we ask which of

them is the power to act—for it seems that this is the office of both. If so, then it might

seem hard to resist the identification of practical reason and the will.

I want to accommodate this thought by insisting that desire is not an extraneous factor

in the explanation of action. In an earlier tradition, represented most famously by

Thomas Aquinas, practical reason and the will are distinguished by their distinct

objects—respectively, the true and the good. But they are nevertheless intimately

related—for practical reason is reason for the sake of action, while the will is the

appetite of reason (1920, Ia q. 79 a. 11; Ia IIae q. 8 a. 1). As the rational appetite, the

will is effectively a part of practical reason; it is the inclination that one has to act in a

certain way, in accordance with one’s thinking. Desire in the relevant sense, then, is not

some extraneous factor like, say, hunger, but is itself an inclination to act in accordance

with thought.

4.9 Conclusion

Rödl says that, since he is writing about “the will—a power—I speak (must speak) …

only of acts of willing that are perfect as willing, that is, of acts that reveal the nature of

the power. It is irrelevant that I may want to do something I ought not to do, or may fail

to be doing what I want to do” (2011, 213). From his perspective, it may look as

though I have tried to build an account of practical reason and the will out of cases

where things go wrong—and this, one might say, is as misguided as it would be to try

to develop an account of theoretical reason out of consideration of cases where the

subject is hallucinating, or is deceived by an evil demon.

In my defense, I would say that my interest has been somewhat limited. I have wanted

to highlight the fact that faults in judgment and faults in motivation are both real and

distinct sorts of fault. One might lose sight of this if one develops an account of

rationality, as for instance Scanlon does, in terms of doing what one takes oneself to

have reason to do; for this might make it look like irrationality is just a matter failing

to do what one takes oneself to have reason to do. True enough, if one would like to

speak that way—the important point, of which I do not want to lose sight, is that there

is more than one way of failing to do what one takes oneself to have reason to do. Both

ways of so failing might be said to be ‘against reason’, but they are against it in

different senses. I take this to indicate that there are distinct aspects of practical

correctness, even in the ‘good case’ where one acts as one should, even when what one

does is, in a sense, utterly singular: one acts on the basis of a good rationale.

Parallel considerations apply concerning the unity of the virtues. Why distinguish the

virtues at all if, as Aristotle argues, they come together? Well, it is only in their perfect

form that they come together. That they come together in their perfect form is, to say

the least, an interesting and provocative thesis. But virtue is rarely, if ever, found in its

perfect form, and the virtues are distinct in those of us who fall short of virtue. We need

to distinguish them, therefore, if we are to know ourselves not just as we might be, but

as we are.

Students also viewed