INTENT‘DNS AN?! CONSEQUENGES:
TOWARD A THEORY OF ART~DESCRIPTEON

fissertation for the Degree of Ph. D.
MLGHIGAN SLATE UNIVERSITY
ARTHUR RONALD MILLER
1975

4-‘-

   
   

LI 3 RA R;
Michigan State

University

 

This is to certify that the

thesis entitled

Intentions and Consequences:

Toward a Theory of Act-Description

presented by

Arthur Ronald Miller

has been accepted towards fulfillment
of the requirements for

Ph . D . Philosophy

 

 

degree in

Major professor

Dstewzfi; 1975

 

0-7 639

I

   

 

‘3' ‘ :
B'NDING BY ‘-

nons & soNS' ‘
suox imam mu.

LIBRARY ewosqs

é :rnmmm. mcmsu

ABSTRACT

INTENTIONS AND CONSEQUENCES: TOWARD
A THEORY OF ACT-DESCRIPTION

BY

Arthur Ronald Miller

This study is devoted to a precise statement of,
and a partial solution to the problem of describing those
bits of behavior which we single out as human actions. I
take as my starting point the claim that any given action
can, in principle, be truly described in an infinite number
of ways. It is this possibility of multiple descriptions
of a given action which gives rise to the sorts of problems
which necessitate the search for an adequate theory of act-
description. It is now a philosophical commonplace that
once we grant the possibility of multiple descriptions of
a given act, we find ourselves faced with a number of dis-
turbing consequences. A given action may be intentional
under one description and yet not intentional under another.
Or we may find that "one and the same action" is sinful or
criminal under one description, commendable or at least

1

Arthur Ronald Miller

exonerable under another, while under yet a third descrip—
tion the act may be of no moral or legal consequence what-
soever. In order to avoid such paradoxes, it is imperative
that a method be devised for selecting from the class of
true descriptions applicable to a given action the "pre-
ferred" or "correct" description, that description which
most appropriately or adequately characterizes the act in
question. This constitutes the problem of act-descriptions.
In Chapter I, after discussing the function of
action-sentences and the difficulties engendered by the
possibility of multiple descriptions of actions, I proceed
to a critical examination of two attempts to specify cri-
teria for selecting the preferred or correct description of
a given action from all of the possible true descriptions
applicable to that action. I first show that each of these
attempts fails hodo the job required. Secondly, it is
argued that all such attempts are, in principle, wrong-
headed from the start, resting as they do on the assumption
that it is possible to specify necessary and sufficient
conditions for a correct act-description. I argue that
this assumption is false, and propose an alternative
analysis of the notion of a correct act-description along

2

Arthur Ronald Miller

the lines of a defeasible concept. That is, I contend that
it is impossible to Specify conditions which are individ-

ually necessary and jointly sufficient for selecting the

 

"preferred" or correct description of an action; the best
one can do is specify minimally necessary conditions and a
set of defeating conditions, that is, conditions the viola-
tion of which is sufficient to defeat a given description's
claim to correctness. According to the analysis suggested
herein, a given description of an action will qualify as a
correct act-description only if it meets the minimally
necessary condition of being true and, moreover, only if
that description is not "misleading." A description (or
any sentence, for that matter) is said to be "misleading"
if it "implies" or gives one's audience to understand things
which are not true or which do not obtain under the circum-
stances. (Clearly, the notion of "implication" employed
here is more akin to Grice's "conversational implication"
as distinct from the formalist's concepts of implication or
entailment.)

In Chapter II this analysis is put to work on the
problem of acts and consequences. The question here is
whether and, if so, to what extent we may correctly

3

Arthur Ronald Miller

redescribe an action solely in terms of one or more of its
consequences. It is shown that, contrary to what some
claim, such redescription is 293 possible in every instance.
First, there are some cases in which a redescription of an
agent's act solely in terms of its consequence(s) fails to
satisfy even the minimally necessary condition of truth; an
explanation is advanced to account for such cases. Secondly,
there are cases in which such redescriptions, while true,
nevertheless fail to be correct insofar as they are mislead-
ing. Five conditions governing the correct redescription of
acts in terms of their consequences are suggested to account
for such cases; these in effect specify the conditions the
violation of which is sufficient for defeating a true de-
scription's claim to correctness. Finally, the implications
of these findings are used to defend the traditional dis-
tinction in ethical theory between deontology and teleology
against those who claim that this distinction is, at bottom,
trivial and unilluminating.

Finally, in Chapter III I discuss (a) the relation
between an agent's knowledge or awareness of what he is
doing (or has done) and the concept of intentional action,
and (b) the bearing of this relation on the problem of

4

Arthur Ronald Miller

act-descriptions. In this context it is argued that wanting
is not a necessary condition for intending. I then divide
act-terms into those which do, and those which do not entail
the presence of intention on the part of the agent, and
argue that all sentences of the simple form ‘A did X'
"imply"-—even if they do not strictly entail--that X was

an intentional action on A's part. This analysis is then
used to explain the misleading nature of a certain class

of act-descriptions which are nevertheless true, and a

final condition is formulated to govern the correctness

of such descriptions.

INTENTIONS AND CONSEQUENCES: TOWARD

A THEORY OF ACT-DESCRIPTION

BY

Arthur Ronald Miller

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Philosophy

1975

To Candi: her perseverance, self-sacrifice, and

encouragement have made this possible.

ii

ACKNOWLEDGMENT S

I wish to acknowledge a debt of gratitude to
Dr. Ronald Suter. Whatever merits may be claimed for
this thesis are traceable, directly or indirectly, to
his useful and timely suggestions, his unyielding
insistence on precision and clarity, and his constant
reminders that "I must mean what I say and say what I
mean." Unfortunately, I cannot share the responsibility
for whatever errors may be contained herein; these are

wholly attributable to me.

iii

TABLE OF CONTENTS

Page

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . 1
Chapter

I. THE NATURE OF ACT-DESCRIPTIONS . . . . . . . . 6

l. The Function of Action-Sentences . . . . . 6

2. 'One Act--Many Descriptions' . . . . . . . 21

3. Correct vs. 'Merely True' Act-

Descriptions . . . . . . . . . . . . . . 33

(i) Rayfield's Proposal . . . . . . 33

(ii) Intention and Convention. . . . 48

4. Description and Purpose. . . . . . . . . . 52

Notes on Chapter I . . . . . . . . . . . . . . 66
II. ACTS AND CONSEQUENCES. . . . . . . . . . . . . 74

l. Redescribing Acts in Terms of

Consequences . . . . . . . . . . . . . . 75
(i) Feinberg's "Accordion Effect"
Thesis. . . . . . . . . . . . 77
(ii) Macklin's "Intervening Agent"
Principle . . . . . . . . . . 86
(iii) True but Incorrect Descrip—

tions . . . . . . . . . . . . 89
(iv) D'Arcy's Elidability Criterion. 97

iv

TABLE OF CONTENTS (cont'd.)

Chapter
(v) A Modified Account of
Elidability . . . . . . .
(vi) Conditions for Redescribing

Acts in Terms of Conse-
quences O O O O O O O O 0

Summary. . . . . . . . . . . . . . . .
2. Implications for Ethical Theory. . . .
Notes on Chapter II. . . . . . . . . . . .

III. AGENT KNOWLEDGE, INTENTION, AND ACT-
DESCRIPTIONS o o o o o o o o o o o o o o

1. Knowing What One is Doing. . . . . . .

2. Bearing on Act-Descriptions. . . . . .

(i) ' Intentionality and Act-Terms.

(ii) Attributing Intentionality.
Conclusion . . . . . . . . . . .
Notes on Chapter III . . . . . . . . .

IV. CONCLUDING REMARKS . . . . . . . . . . . .

BIBLIWRMHY O O O O O O O O O O O O O O O O O O O

Page

110

118
135
137

148

153
154
185

192
203

211
215

220

226

INT RODUCT I ON

The topic of human action has been discussed
extensively in the recent literature. The problem of
identifying what sorts of items of human behavior are to
be counted as actions and why they are to be so counted,
as well as the problems of individuating and explaining
human acts have come in for most of the discussion. The
specific topic I wish to address in this dissertation is
the problem of how we are to describe those bits of be-
havior which have been singled out as actions. J. L.
Austin was largely responsible for formulating and setting
the question as we are now familiar with it. In "A Plea
for Excuses," Austin writes,

To do so [i.e. to state explicitly what is
being excused] is all the more vital because
it is in principle always open to us, along
various lines, to describe or refer to "what
I did" in so many different ways. This is
altogether too large a theme to elaborate
here. Apart from the more general and ob-
vious problems of the use of "tendentious"
descriptive terms, there are many special
problems in the particular case of "actions."

Should we say, are we saying, that he took
her money, or that he robbed her? That he

 

l

knocked a ball into a hole, or that he sank
a putt? That he said "Done," or that he
accepted an offer? How far, that is, are
motives, intentions, and conventions to be
part of the description of actions.*

This feature of human action to which Austin draws
our attention has come to be regarded by many philosophers
as quite significant: namely, that one and the same action
may, in principle, be truly described in innumerable ways.
For it is this possibility of multiple descriptions of a
single action which gives rise to a number of problems which
many find quite disturbing. We will come to see that it is
not a matter of indifference which of several true act-
descriptions we choose to characterize a given item of be-
havior. On the contrary, varying the act-description can
sometimes affect a whole host of other features peculiar to
human acts in which we are particularly interested. For
example, given an item of behavior, under description D1
we may hold the agent accountable for his act and perhaps
blame him for it, whereas under another (true) description,

D2, his act might not even qualify as the sort of thing for

which persons are praised or blamed at all. It is problems

 

*J. L. Austin, "A Plea for Excuses," in Philosophical
Papers, ed. by G. Warnock and J. O. Urmson (Oxford:
Clarendon Press, 1962), p. 148.

 

such as these which emphasize the need for a method of
choosing a "preferred" description of a given act; that is,
what is needed is some method for selecting from the class
of true descriptions of a given act, that description (or
descriptions) which most apprOpriately characterizes or
describes the act in question. And it is this need which

I hope to satisfy--at least partially--during the course
of this study.

In Chapter I, after discussing the function of
action-sentences, I attempt to indicate-and spell out in
detail the sorts of problems engendered by the possibility
of multiple descriptions of a single action and show how
these bear importantly on other issues in ethics, philOSOphy
of mind, and philosophy of law. I then proceed to an expo-
sition and critical analysis of two suggested methods for
selecting the "preferred" or "correct" description of an
action from the larger class of true descriptions applicable
to a given act; these in effect constitute attempts to spe-
cify necessary and sufficient conditions for distinguishing
"correct" descriptions of human actions from those which are
"merely true," so to Speak. It is argued that these at-

tempts fail and, finally, an alternative analysis is

sketched out, an analysis which, I hope, does the job re-
quired of it while avoiding the shortcomings of the other
analyses. I then proceed in Chapters II and III to fill in
the details of this analysis and put it to work on the
problem of act-descriptions as it relates to the notions

of consequences and intention.

There are a couple of qualifications that I should
make clear here at the outset. I will be concerned here
only with non-linguistic human acts. The topic of speech-
acts is altogether too extensive to be included within the
scope of this inquiry. Many of the things I have to say
about the description of non-linguistic human acts may or
may not apply to speech-acts as well. But this is a sub-
ject for another study. Secondly, I will restrict my dis-
cussion to those acts whose agents are individuals. It is
clear that there are often cases in which groups of indi-
viduals act collectively. Such groups may consist of as
few as two members, or they may be as large as the aggregate
of all the constituents of a community. Juries convict and
acquit defendants, political parties endorse and reject
platforms, tribes ostracize members who have fallen from

favor, and so on. But to pursue this issue would invariably

lead us into an investigation of many other problems such
as collective responsibility, problems which would take us

far afield of the proposed tOpic.

CHAPTER I

THE NATURE OF ACT-DESCRIPTIONS

l. The Function of Action—Sentences

 

What are we doing when we say that so-and-so did
such-and-such? What am I doing when I say, for example,
that 'Houdini escaped his bonds,‘ that 'Joan is buying a
loaf of bread,‘ that 'Jones killed his mother,‘ or that
'I am playing badminton'? Am I simply reporting what the
agent did or is engaged in doing, am I passing judgment on
his action, am I somehow ascribing responsibility to the
agent for his action; or may it be that I am doing a number
of these things simultaneously? The first question we must
examine is what sorts of jobs are or may be performed when
I employ a locution of the form 'A did X.‘ This question
is to be kept distinct from a separate, related question,

namely that of the logical form of such sentences. Our

 

question concerns the function of such sentences, not the
status of the sentence the speaker utters or the proposi-
tion expressed by that sentence.

6

This question becomes important in light of the
fact that some notable phiIOSOphers have held that when
one says something of the form 'A did X,‘ this sentence
93323 serves to describe an action. Thus before it is
possible to inquire into the nature of act—descriptions,
it will first be necessary to show that such locutions can

and, in fact, often do function as bona fide descriptions.

 

What, then, is the function of action-sentences of
the form 'A did X' or 'A is x-ing'? One answer to this
question which has gained some currency is that sentences

of this form serve to ascribe responsibility for the act

 

to the agent in question. Furthermore, it is implied, such
sentences function only to ascribe responsibility to an
agent for the commission (or omission) of the act; they
23325 serve to describe the act in question. H. L. A. Hart
and A. I. Melden are two cases in point. Negatively, both
Hart and Melden are concerned to dispute traditional
analyses which break an action down into two components,

a bodily movement plus some antecedent or concurrent mental
event. In arguing against such analyses, both are at pains
to point out that such accounts fail to capture many of the

essential features of human action. Hart argues that the

traditional naive views of action err in thinking that an
adequate analysis of the concept of a human action can be
given exclusively in the form of some combination of de-
scriptive statements. For Hart, on the contrary, actions
are events to which we ascribe rights and responsibilities.
Melden for his part argues that if action sentences were
descriptive, "then in order that I might know what I was
doing in any given case, I would need to know what bodily
movements took place, and this I could know only by ob-

. l .
serVing my own movements." Apart from the non sequiter

 

involved in Melden's argument,2 the important-thing to
notice in both cases is the subtle shift in meaning Hart
and Melden have introduced into the notion of 'description.‘
As Yolton observes, both "seem to use 'description' as
applied to physical behavior only."3 This becomes clear
in another place where Melden appears to concede that
perhaps we are not ascribing responsibility in some cases
in which we say of so-and-so that he did such-and-such:

There are cases, of course, in which sentences

are employed in describing the behavior of our

fellows and in which there is no ascription of

responsibility. I have already mentioned the

language of coroners and physiologists, in

which a position of neutrality is taken with

respect to the responsibility of the indi—
vidual.4

 

This sort of antipathy to the descriptive function of
action-sentences (or, more precisely, the denial that such
sentences can or do perform this function), it seems, leads
Hart and Melden to a one-sided notion of action. "Quite

clearly," as Yolton observes, "an action for these writers

T

 

is a happening that has mgral significance."5 As Yolton
declares elsewhere, "for both Hart and Melden, the term
'action' is used in the moral sense; they are trying to
explicate the meaning of an action in such a way that some
light will be thrown upon the moral problems of responsi—
bility and right."6 By focusing exclusively on the moral
and legal features attending some human actions and the
correlative ascriptive function of action-sentences, Hart
and Melden, like the predecessors they so vehemently oppose,
seem to be just as guilty of ignoring other important fea-
tures of human actions. The mistake here seems to be that
of supposing that since "the domain of morality is the
domain of action," as Hampshire puts it,7 the two domains
are therefore coextensive.

To show why this conception of the function of
action-sentences is wrong-headed, it is useful to distin-

guish (at least) two senses of 'responsibility.‘ On the

10

one hand, A may be said to be 'responsible' for x in the
sense that A, in fact, did X; that is, A is (was) the
author of x, X is to be ascribed to A as its doer or agent.
On the other hand, A may be said to be 'responsible' for X
in the sense of being morally or legally accountable; that
is, A may be held liable for the commission (or omission)
of X. These two senses, though related, must be kept dis-
tinct. It seems clear that in analyzing action as funda-
mentally a moral concept, Hart and Melden-have in mind this
second sense of 'responsibility,‘ namely that of moral/legal
liability or accountability. It is a peculiar feature of
distinctively human actions that we are often interested in
their specification for the sake of fixing responsibility
for moral or legal assessment. But from this obviously
true observation, we must not be tempted to conclude that
the gnly way in which sentences of the form 'A did x' or

'A is X-ing' function is to assign responsibility for the
act in the sense of accountability. To show that this is
so, I will argue that neither sense of 'responsible' dis-
tinguished above entails the other, though in many (perhaps

even most) cases they go hand in hand.

11

For the sake of simplicity, I shall call the first
sense of 'responsible,‘ namely the sense in which the act X

is ascribed to A as its agent, causal responsibility; and I

 

will refer to the second sense, the sense in which A is
held morally or legally liable or accountable for the com—

mission (or omission) of X moral responsibility. In the

 

first place, A may be held morally responsible for an act
which he did not (or, in some cases, logically could not)
commit. We may hold Richard Nixon morally responsible for
the break-in at the Democratic National Committee in the
Watergate, even though he did not take part in the burglary
and was not even present at the scene. As an instance of a
case in which I may be held morally responsible for an act

which I am logically precluded from performing, consider

 

the following: I may, through my acts of cruelty, negli-
gence, and so on, be held morally responsible for my wife's
suicide; I may legitimately be blamed for it and held
liable by my fellows. Nevertheless, her suicide is not
something I did; being suicide, only she could do it.

These matters will be taken up in further detail when we
come to discuss the notion of consequences; at this point

I am only trying to show that responsibilty in the sense

12

of moral responsibility does not entail that the person
held thus responsible is necessarily the author of the act
in question.

The converse of this claim provides the basis for
the second objection to the Hart-Melden View that action-
sentences are primarily (or exclusively) ascriptive. A
person may be the author of an act (i.e. be causally re-
sponsible) and yet not be held morally responsible at all
for its execution. This is so for a very simple reason,
namely, that many of the things we do simply are not can-
didates for praise or blame at all, and consequently we are
not held (morally) responsible for them; they are, so to
speak, morally "neutral"; and interest in these are not, as
Melden claims, restricted to coroners and physiologists.

If I stretch my arms or walk to the kitchen, I am certainly
'responsible' for these acts in the sense that I am their
author, but the question of moral responsibility or ac—
countability does not even arise (excluding extraordinary
circumstances, of course). As Geach remarks,

as regards hundreds of our voluntary or in—

tentional acts, it would in fact be absurdly

solemn, not to say melodramatic, to talk of

imputation and exoneration and excuse, or

for that matter of praise or reward.
Ascribing an act to an agent just does not

13

in general mean taking up a quasi-legal or
quasi-moral attitude, and only a bad choice
of examples could make one think otherwise.
(As Wittgenstein said, when put on an un-
balanced diet of examples ghilosophy suffers
from deficiency diseases.)

 

If what I have argued is correct, then the Hart-Melden View
that action-sentences do not describe but serve only to
ascribe responsibility would appear to be unsound as a
general account of the function of action—sentences. Their
mistake, as I have tried to suggest, lay in overextending a
fruitful analogy between certain types of human actions and
moral and legal concepts in trying to account for the con—
cept of action in general. As Yolton remarks, "the expli-
cation of moral terms by reference to legal terms should
not be allowed to absorb the moral into the legal. Simi-
larly, the philosophical analysis of a concept like action
should not be identified with the practical talk of making
moral judgments."9

Thus far I have only tried to argue that the Hart—
Melden View of the function of action-sentences will not do
as a general account. I would like now to turn to a posi—
tive account of the function of such sentences, including
a discussion of what I take to be right in the Hart-Melden

analysis. First a general point about the use (i.e.

14

function) of sentences. There is no-reason to conclude
from the fact that a sentence is used to perform any par-
ticular illocutionary act, that it can be used to perform
only that act. There is nothing essential about any given
sentence which dictates how it can be used. Most sentences
can (typically) be used to perform a variety of tasks. For
example, I may use the sentence 'The window will be closed'
on one occasion to make a prediction concerning an imminent
state of affairs; on another occasion I may use the same
sentence to issue a command to a subordinate or mine within
hearing distance; this sentence is not necessarily either a
prediction gr'a command, but rather may be used in either
way.10 Similarly in the case of action-sentences, even if
it were the case, as Hart and Melden claim, that such sen-
tences serve to ascribe responsibility, it would not follow
that this is all they do (or can do). With this in mind
let us now examine some of the ways in which such sentences
may function.

We may begin by making a rather elementary observa-
tion. Any sentence of the form 'A did X' or 'A is X-ing'
does, if true, serve to pick out the agent in question. To

the extent that this is the case, we may say that such

15

sentences function to ascribe what I have termedwcausal re-
sponsibility to the agent, that is, they serve to identify
the author of the act. But to state the matter in this way
is misleading. The function of ascribing causal responsi-
bility to an agent is better limited to those cases in
which we are genuinely interested in identifying the author
of the act in question. If we know that the window has
been broken but have no idea of who might have broken it,
then the sentence 'Jones broke the window' may properly be
said to identify the agent. If, on the other hand, we know
that Jones broke the window, but what we are particularly
interested in is how he broke it, then the sentence 'Jones
broke the window with a stone,‘ while it does serve to pick
out the agent, is not expressly designed for this purpose.
Given that we already know that it was-Jones who did it,
the sentence '§g_broke it with a stone' would have done
just as well, even though this sentence itself serves to
pick out no one in particular (since this sentence contains
the indexical 'he,‘ it may be used on various occasions by
different speakers to refer to various individuals; of
course, the context of the utterance will usually make it

clear which individual is being referred to). In other

16

words, we can distinguish the illocutionary act performed
by the speaker who utters a sentence of the form 'A did X'
from the function(s) that the sentence itself may serve,
even when these happen to be the same.11 We may say, then,
that although it is the case that every sentence of the
form 'A did X' does, if true, function to pick out the
agent in question, it does not follow that every time a
sentence of this form is uttered the speaker is performing
the illocutionary act of ascribing (causal) responsibility;
though, as we have just seen, there are cases in which this
lg precisely what he is doing.

The broken window example is helpful in distinguish-
ing yet another function of action-sentences. Many actions
are such that the agent is held accountable for their com-
mission or omission. And here I am quite willing to concede
what Hart and Melden want, namely, that many sentences of

the form 'A did X' or 'A is X-ing' do serve inter alia to

 

ascribe moral or quasi-moral responsibility to the agent.
If it is truly said that 'Jones broke the window,‘ then
Jones is held accountable for his action; some explanation,
excuse, disclaimer, justification, candid admission of

guilt, or whatever is called for on the part of Jones.

17

Here again we can distinguish the functions of the sen-
tence, the job it serves to do, from the illocutionary act
performed by the speaker. Here it is especially important
to note that if a sentence of the form 'A did X' or 'A
X-ed' is used to ascribe liability (or what we have been

calling moral responsibility), then that sentence a fortiori

 

functions to ascribe causal responsibility in the sense of
picking out the author of the act.12 Thus, even if the
speaker is explicitly performing the illocutionary act of
ascribing moral responsibility in saying that 'Jones broke
the window,‘ his utterance of the sentence functions never-
theless to pick out or identify the author-of the act (viz.
Jones). Of course the speaker may, depending on the con-
text and his intentions, be performing both of these acts,
in which case we have an instance of a sentence being used
to perform two jobs at the same time.

In order to isolate yet another function of some
action sentences, consider instead of the relatively in-
nocuous act of breaking a window, the case of an action to
which great moral/legal significance is attached, say
murder. Assuming (and it seems reasonable to suppose)

that we all share a sense of outrage and disapprobation

18

of murder, it would seemrplausible to suggest that a sen-
tence like 'Jones murdered his daughter in cold blood' may
function to express the speaker's condemnation of Jones for
his act of murder. Here again we see a case of an action
sentence performing a number of functions, any one or all
of which may be the speaker's intended use; in the above
sentence not only is Jones identified as the murderer (as-
cription of causal responsibility) and held liable for its
commission (ascription of moral responsibility), but the
sentence also expresses the speaker's-moral-assessment of
the action. Far from any of these functions excluding the
others, here is a paradigm case in which an action sentence
serves to function in a number of distinct ways simulta-
neously.

Now what can be said of the descriptive function of
such sentences? What is common to each of the examples we
have considered is that in each case an act or an act-type
has been mentioned. Once again it is useful to distinguish
what a speaker is doing in uttering a sentence from the
function(s) served by the sentence he utters. In the case
of the ascription of causal responsibility I remarked that

the performance of this illocutionary act is best limited

19

to those cases in which we are specifically interested in
identifying the author of the act in question. Similarly
in the case of description we may, if we like, restrict the
illocutionary act of describing (an action) to those cases
in which we are interested not in identifying the author of
an act, but rather in discovering what it is that the agent
allegedly did. For example, were I to ask not 'Who did
such-and-such?‘ but 'What did Jones do?‘ and you were to
answer that 'Jones broke the window,‘ we-may say that you
were performing the illocutionary act of describing (an
action). This feature of action sentences which we have
noted, namely, that they may be used to perform various
illocutionary acts, is a function of what Feinberg refers
to as their "contextual presupposition." In this sense,
what illocutionary act is being performed by a person who
utters a sentence of the form 'A did X' will be determined
largely by the interests and purposes of the speaker and
his audience.

What is not an ascription in one context may

well be so in another, depending on the con-

cerns of the speaker. If the question is

'What did Jones do?‘ then the sentence ‘Jones

did A' describes what Jones did; but if the

question is 'Who did A?‘ then the sentence
'Jones did A' ascribes A to Jones.13

 

20

It nevertheless remains true that, regardless of the
speaker's intention or the illocutionary act he performs,
the sentence he utters does serve to pick out, specify or,
in other words, describe what the agent did. (Of course
one may explicitly be performing the illocutionary act of
describing an action, in which case we have a clear in-
stance of an act-description.) Just as in the case of the
illocutionary act of ascribing moral responsibility in

which the sentence 'A did X' also functions a fortiori to

 

identify the author of the act, so too whenever a sentence
of this form is uttered, an action is being described or
identified (by the word or phrase which is substituted for
'X'), in spite of whatever else the speaker is doing. If
we take our original catalogue of action-sentences, namely,
'Houdini escaped his bonds,’ 'Joan is buying a loaf of
bread,‘ 'Jones killed his mother,‘ and 'I am playing bad-
minton,‘ the phrases which serve as the descriptions of
these actions are 'escaped his bonds,‘ 'buying a loaf of
bread,‘ 'killed his mother,‘ and 'playing badminton,‘ re-
spectively. And it is in this relatively obvious sense

of an act-description that we are interested.

21

2. One Act--Many Descriptions

In his excellent little book Human Acts: An Essay

 

in their Moral Evaluation, Eric D'Arcy begins by setting
forth and discussing three claims concerning act—
descriptions which he elevates to the status of theses.
We will have occasion to discuss each of these in detail
during the course of our inquiry. For our present pur—
poses, however, I would like to focus on D'Arcy's first

claim:

Thesis One: There is not necessarily one,

and only one, correct description of a
given act.12f

 

 

 

The view expressed here by D'Arcy is one which has come to
be commonly held, namely, that not only do sentences of the
form 'A did X' in fact serve to describe actions, but that
any given action may be truly described by many different
action-sentences of this form. This claim, which I shall
refer to as the 'one act--many descriptions' thesis, has
been endorsed by nearly all of those who have considered
the nature of act-descriptions in recent years. We have
already had occasion to quote Austin in this connection:

u

. . . it is in principle always open to us, along various

lines, to describe or refer to 'what I did' in so many

22

different ways."15 Donnellan has referred to this feature
of act-descriptions as a "well known fact."16 O'Shaughnessy
speaks of the permanent possibility of redescription in his

discussion of human action.17 Stuart Hampshire makes this

point explicitly in his now-famous Thought and Action:

 

Even when he is wholly absorbed in something
that he himself would call a single activity,
and when his attention is in‘no-way-divided,
there is necessarily a variety of true ac-
counts that could in principle be offered of
this single activity.18

And the list goes on and on.19
This possibility of multiple descriptions of a
single action is not itself surprising for, as Hampshire
points out, "at any moment in a man's waking and conscious
life there is always a set of possible true answers to the
question—-'What is he doing now?”20 For example, we may
have occasion to describe the action of a quarterback in a
football game variously as 'running the option play,‘
'throwing a pass,‘ or 'leading his team,‘ all of which may
be true descriptions of what he is doing at a given time.
Or to take a case from D'Arcy, "we might think of~a test-
pilot who," to the question 'What are-you doing now?,'

"could truthfully answer, 'Holding the joy-stick'; 'Look-

ing through the windscreen'; 'Listening to the

23

control-tower'; 'Smoking a cigarette'; . . . 'Flying over

'"21 Not

the North Sea'" or "'Testing the new Vickers jet.
only may we vary the description of yaaa the agent is doing
in this way, but any given description may be expanded to
mention various features which we might wish to include in
our characterization of what the agent did. We may wish,
for example, to include reference to the time at which the
act is performed or the means employed in performing it or
the manner in which it was executed. Rescher-cites all

sorts of features which may be included in-"what might be

called the canonical description of an action“-in his

 

"catalogue of the keylgeneric elements of actions"; these

 

include such things as the manner in which an agent did

what he did, the means he employed-in doing it, when, where,

 

and under what circumstances it was done, what caused him

 

to do it, with what aim he did it, in what state of mind

 

he did it, and so on.22 Thus, for example, we may say
simply that 'A did X,‘ or we may say that 'A did X at

time t' or that 'A did X at time t by employing means M,‘
or so on, in each instance truly describing the-action in
question.23 In short, not only may we offer different de-

scriptions of an act by referring differently to what the

24

agent did, but even granting that what he did was«X; we may
still vary the description by including more'or fewer of
these "key generic elements of action" in the sentence we
choose to describe the act. If all this is true, then it
appears that the 'one act--many descriptions‘ thesis is
indeed well-founded; of any given action, one-may always in
principle come up with any number of sentences truly de-
scribing what the agent did or is doing.

As I mentioned above, it is not the 'one act—~many
descriptions' thesis itself which is alarming; rather, it
is the implications of this thesis which generate the prob-
lems and difficulties which constitute the-point of depar-
ture for my inquiry. Some of these implications are far-
reaching and of the utmost consequence.’ I would like to
take this opportunity to indicate what some of these prob-
lems are before going on to examine suggested methods of
dealing with them.

If one reflects for a moment, it becomes clear that
there is a sense in which a single act-type may,-depending
on the circumstances and context, constitute doing different
things. In signing one's name, for example, one may be

giving his autograph to an admiring fan or endorsing a

25

check. Furthermore, and more importantly, there may be
cases in which, again depending on the context, a single
act-type may constitute doing things which are inva sense
"incompatible." Signing one's name may, in one context,
be to issue a reprieve while, in a different set of cir-
cumstances, the "very same act" may constitute issuing a
death warrant. In other words, as D'Arcy remarks, "circum—
stances may affect not only one's final moral evaluation
of an act, but also one's characterization of it: in some
circumstances X is P, in others it is Q."24
A couple of examples may help to illustrate the
bearing of this possibility of multiple act-descriptions
on problems associated with ethics and morality. Consider,
for instance, a case from Kant. Suppose a man tells a lie
in order to save a friend's life. It seems that we may
truly describe this man's act variously as 'telling a lie'
or 'saving a life.‘ Now certainly these two descriptions
are "incompatible"--not in the sense that doing either of

these precludes the possibility of doing the other--but in

the sense in which there is a prima facie presumption

 

against lying whereas there is a strong presumption la

favor of saving lives. Which of these descriptions we

26

choose, then, may have a tremendous influence-on our moral
assessment of the man's action and his character.25
Another example may serve to illustrate even more
acutely the sorts of problems for morality engendered by
the 'one act--many descriptions' thesis. Consider the ac-
tivities of Adolph Eichmann on a given day in 1943. I and
many others are prepared to say of Eichmann that 'he exe-
cuted many innocent German Jews' or even that 'he committed
genocide,‘ a crime as heinous as any the human imagination
can conceive. How might Eichmann himself and his col-
leagues, on the other hand, be apt to characterize his
activities? We can easily envisage Eichmann describing
the activities under consideration as 'obeying his legi—

timate superiors' or perhaps even 'doing his duty.” And

it seems as if there is nothing at least prima facie wrong

 

in obeying one's superiors, at least not in the sense in
which there is a presumption against it in the way that
there is a presumption against, say, committing genocide.
And after all it is true to say of Eichmann that he was
obeying his superiors--though it is not at all clear that
it is true to say that he was doing his duty. Now consider

a third party, oblivious of the circumstances, consequences,

27

and so on of Eichmann's actions. This-person'may feel
quite content to describe what Eichmann did-as ‘activating
an oven's firing mechanism'; and this description, too, we
may suppose is true. This implication of the 'one act--
many descriptions' thesis is disturbing indeed. One and
the same action may be sinful or criminal under one de-
scription ('executing innocent German Jews'), exonerable
or perhaps even commendable under another ('obeying one's
superiors'), while under still another description the act
is not even subject to moral or legal praise or censure
('activating an oven's firing mechanism').

The example just considered suggests-further prob-
lems associated with describing actions. Let us assume for
the moment that Eichmann himself did not press the button
to activate the oven's firing mechanism, but that instead
he gave the order to one of his subordinates to do so.
First we have Eichmann's act of ordering his subordinate
(let us call him Schultz), the consequence of which is the
firing of the incinerator; Schultz's act, in turn, results
in the death of the innocent Jewish victims. Here we have
at least two questions to consider. First, can we distin-

guish an act from the consequences of that act and, if so,

28

how? Is it merely a practical, arbitrary distinction we
commonly make, or are there any logical criteria for dis-
tinguishing an act from its consequences? Secondly, to
what extent, if any, can we correctly redescribe an action
in terms of its consequences? With respect to the present
case, though we may feel no hesitation in holding Eichmann
morally responsible for the consequences of his act (viz.
the deaths of the Jewish victims), can we substitute
'Eichmann' for 'A' intfluasentence-schema 'A killed 150
innocent Jews' to obtain a true sentence? Once again, are
there any logical or quasi-logical criteria for distin-
guishing cases in which we may correctly redescribe an act
in terms of its consequences from those in which such re-
description is precluded as a genuine possibility? I will
come back to address these and related problems in Chap-
ter II; at this point I am only concerned to indicate the
sorts of problems which must be dealt with.

Returning again to our original illustration, we
might ask what sorts of descriptions genuinely count as
descriptions of acts or act-types, and which are more
properly characterized as descriptive of goals, endeavors,

plans of action, and so forth. There does appear to be an

29

element of particularity associated with the notion of aa
action. Murder, telling a lie, and pitching a ball are
relatively clear cases of act-types. It also seems plaus-
ible to suggest that leading a good life or conquering the
world, on the other hand, are not properly termed actions
in any but a quite extended sense. So in the Eichmann il-
lustration, for example, it is not clear that the phrase
'committing genocide' can properly be regarded as referring
to or describing an act or an act-type. Perhaps this is
part of what Austin has in mind in Speaking of various
stages or phases of action. Committing genocide, unlike
killing a single innocent person, cannot be accomplished
in a temporal stretch sufficiently restricted to refer to
it as aa action.26

In addition to its bearing on moral and legal
issues, the problem of describing actions is also inex-
tricably bound up with wider issues in philosophical psy-
chology. Consider, for example, Davidson's now-famous
illustration27 of the man who flips the electric switch,
turns on the light, and illuminates the room; unbeknownst

to him, however, he has also succeeded in alerting a

would-be burglar that he is now at home. We may, it seems,

30

truly describe the man's action, what he did, in a variety

 

of ways, including the following:

(1) He turned on the light.
(2) He alerted a prowler.

But whereas the man had "every intention,‘ so to speak, to
turn on the light, he had no intention whatsoever to alert
the prowler; indeed, he was not even aware that there was a
prowler inside. From this it would seem to follow that a
given action may be intentional under one description, and
yet not intentional under another. Yet it seems that we
are dealing here with one and-the same event.

These considerations among others have led some
philosophers to complain that such analyses make too much
depend on the descriptions we proffer for our own and
others' actions. Arthur Cody has succinctly voiced the
unrest felt by many of these philosophers. The picture
suggested by the 'one act--many descriptions' thesis, he
claims, "yields a disturbing paradox." "Can there," he
asks, "be justice in our praise or blame when everything
depends on which description we select to judge a man's
actions under? It seems there cannot if many descriptions

are applicable and if all applicable descriptions are,

though different, true."28 Given the sorts of difficulties

31

engendered by the 'one act--many descriptions' thesis, are
we forced to simply throw up our hands in desperation, or
may it perhaps be possible to formulate some guidelines for
choosing from the multitude of true descriptions of a
single act one which is the "preferred" description? A few
philosophers, notably Cody, have gone so far as to deny the
'one act--many descriptions' thesis, claiming instead that
one and only one description is applicable to any given
action. Others, however, accept the conclusion which Cody
bemoans, namely that much aaaa indeed depend on the descrip-
tions we give of actions, but have set themselves the task
of elucidating the notion of action-description and at-
tempting to solve the sorts of difficulties and dispel the
sorts of paradoxes to which Cody draws our attention. It
is with these latter philosophers that I plan to join
forces.

' What is needed, then, is an analysis which will
provide some criteria or guidelines for marking off one or
more descriptions of a given action as the "preferred" de-
scription(s); that description which, among the many pos-
sible true descriptions, most appropriately or correctly

characterizes the act in the given circumstances. We may

32

find that any attempt to-formulate-a rigid set of necessary
and sufficient conditions for "preferred" or "correct" act-
descriptions is destined to fail to do the-job required of
it. Concepts like propriety and correctness are, after
all, highly dependent on human interests, purposes, conven-
tions, and so on. Whether or not a sentence is true depends
on whether or not it satisfies more or less easily speci-
fiable truth-conditions. The concept of propriety, on the
other hand, does not lend itself to an analogous explica-
tion. But from the fact that it may be impossible to
formulate a set of necessary and sufficient conditions for
"correct" or "appropriate" act-descriptions, and from the
observation that such concepts are dependent on human pur-
poses and conventions, it does not follow that there are

or can be no considerations to guide us in our inquiry.

The rest oftmis study will be devoted to an attempt to

determine and explicitly formulate such considerations.

33

3. Correct vs. 'Merely;True'
Act-Description529

 

 

It now appears to be a philosophical commonplace
that, once we grant that a given item of human behavior may
be truly described in any number of distinct ways, we find
ourselves faced with the sorts of difficulties and para—
doxes discussed in the last section. What is sorely needed,
then, is an analysis which will provide some criteria,
however inexact, for selecting from the class of true de-
scriptions of an action that descriptionr(or those de-
scriptions) which most adequately,-appr0priately, or 'cor-
rectly' characterizes or identifies what the agent is doing
at any given moment. In this section, I would like to
examine in some detail two such proposed analyses for dis-
tinguishing the 'correct' description(s) of a given action
from the much larger class of true descriptions applicable

in a given case.

(i) Rayfield's Proposal

 

The first of these analyses was first proposed by
David Rayfield in a paper entitled "On Describing Actionsfl30

To the best of my knowledge, this paper represents the only

34

explicit attempt to date to formulate-a set of necessary

and sufficient conditions for distinguishing"correct' from

"merely true" act-descriptions; furthermore, it is expressly

designed to meet the pressing need in action theory engen-

dered by the 'one act--many descriptions' thesis. I would

now like to turn to an examination of the proposal. First

I will attempt to present a clear and accurate exposition

of the analysis itself. I will then go on to argue that

the analysis fails to do the job required of it for two

reasons. First, I will try to Show that Rayfield‘s analysis

of true descriptions is circular insofar as it turns on his

notion of an 'unbound action.‘ Secondly, and independent

of the charge of circularity, I will argue that the basis

upon which Rayfield draws the true-correct distinction

leads to certain unacceptable consequences for action

theory. But more of this later--first the analysis itself.
Rayfield proposes a threefold distinction between

various descriptions of a single action. First there is a

distinction between descriptions which are applicable to

an item of an agent's behavior and those which are not

applicable. A description Dn is said to be applicable to

 

. . . . . ' 31
A's behav1or "if and only if Dn 15 either true or-correctfl'

35

Thus, he introduces a twofold distinction within the class
of applicable descriptions, namely, the distinction between
descriptions which are EEES and those which are correct.
He defines these concepts as follows:

A description Dn, of A's behavior, X, is true

if and only if A did do what he is said to
have done in Dn.

 

A description Dn is correct if and only if:
(i) Dn is true; and (ii), Dn is the same as
Dn', Dn' being the description of x under
which A X's, i.e. A's own description of
his behavior.32

A description which is not applicable is false. A descrip-

tion which is correct is a fortiori true, but not con-

 

versely. Truth does not ensure correctness, though cor-
rectness ensures truth, correctness being a stronger con-
dition than truth and including it as a necessary condition
of itself.

Before examining the implications of this analysis,
it may be helpful to see it at work as applied to an illus-
tration offered by Rayfield.33 Suppose a fireman in a
power shed has certain duties, among which are keeping the
boilers fired, checking the chlorine content of the water
in the hot water system, and so on. In the midst of one of

his meter readings a fire breaks out and the fireman Sets

36

himself to extinguishing it. Rayfield offers several de-
scriptions which might be given of the fireman's behavior,
including the following:

(1) He's fighting the fire.

(2) He's praying.

(3) He's swearing.

(4) He's reducing a supply of sand (under the
sign 'FIRE').

(5) He's neglecting his duties (e.g. the forgotten
chlorine reading).

(6) He's firing the boiler.
In the first place, (6) is simply false; it is not a true
statement of what the fireman-is doing. Hence description
(6) is not applicable. (1)-(5) we may accept as true and
hence applicable. But which of these (1)-(5) are correct
as well as applicable? By appealing to the criteria for
correct descriptions we may conclude that those and only
those descriptions are correct for which the fireman him-
self would give a candid, affirmative response to the
question 'Were you (are you) X-ing?‘34 For example, were
we to ask the fireman 'Were you praying?‘ and were he to
candidly answer 'Yes,‘ we could then take description (2)

as correct as well as true. If, on the other hand, we

37

were to ask him 'Were you (are you) neglecting'your‘duties?'
and were he to answer 'Hell no!,' description (5) while
true (it truly describes what the fireman was doing or,
more precisely, what he failed to do), it would neverthe—
less fail to meet the other necessary condition for quali-
fying as a correct description. Thus while both descrip-
tions (2) and (5) are true and hence applicable, only de-
scription (2) is correct. Rayfield acknowledges that there
may, on a given occasion, be more than one correct descrip—
tion of an agent's act, namely when the agent responds
affirmatively to the question 'Are you X-ing?‘ or 'Did
you X?’ under two or more descriptions of his action;

nevertheless, claims Rayfield, "it may be generally the

 

case that one performs an action under only one descrip-
tion."35
So much then for Rayfield's analysis. Is there
anything about this proposal which should make us hesitant
to accept it wholesale? Leaving the applicable-inapplicable
distinction aside for the moment, there is something odd in
Rayfield's subdivision of applicable descriptions into

those which are "merely true," so to speak, and those which

are not only true but correct as well. More specifically,

38

it is the basis upon which Rayfield draws the distinction
which should make one suspicious. It is not clear exactly
what job he expects the true-correct distinction to per-
form, but it is illuminating to discover what sorts of
claims one finds himself committed to if he does accept
this distinction the way he marks it.

Let us focus for a moment on what Rayfield has to
say concerning those descriptions which he terms true but
not correct. He tells us in one place that "true descrip-
tions will be descriptions of items of A's behavior which
qualify; not as bound actions, but as either mere doings or
unbound actions."36 "Mere doings" are items of behavior
such that no agent could, on any occasion, decide to per-
form. Rayfield gives the example of sneezing; one could
never decide to sneeze, even though one could decide to
undertake measures conducive to sneezing (say sniffing

pepper). Mere doings can never qualify as bona fide human

 

acts, bound or unbound. What is important for our purposes
is the concept of an unbound action, for it is to such
actions that true--but not correct--descriptions apply. To
discover what is meant by the notion of an unbound action

we must first attend to Rayfield's analysis of~a bound

39

action, for the former are defined negatively-in-terms of
the latter. Rayfield specifies four jointly necessary and
sufficient conditions which a given item of behavior X must
satisfy to qualify as A's bound action. These may be sum-
. 37
marized as follows:
(1) X must be a doing (as opposed to a suf-
fering, or something which happens to
A).
(2) X must be decidable; that is, X must be
such that someone, not necessarily A,
could on some occasion, not necessarily
this one, decide to X.
(3) A must be responsible for X.
(4) A would give, if he were asked it, a
. candid, affirmative answer to the ques-
tion 'Are you X-ing?‘ or 'Did you X?‘
Only items of behavior which successfully meet all four
conditions qualify as bound actions. A given item of be-
havior which fails to meet condition (2) counts as a mere
doing. An item of behavior which satisfies conditions
(1)-(3) but which fails to meet condition (4) qualifies
as an unbound action. But, alas, condition (4) which
serves to distinguish bound from unbound actions is pre-
cisely the criterion which distinguishes correct from true

descriptions of actions. Thus it seems as if-we have come

full circle. Our original task was to get clear on the

40

notion of a true description; to do-this we found it neces-
sary to explicate the concept of an unbound action; But it
turns out that the concept of an unbOund action is analyzed
in terms of true descriptions. In short, this line of
questioning seems to have gotten us nowhere.

I think the crux of the problem will become ap-
parent if we examine the force of condition (4) of bound
actions which, as we have seen, also serves as the cri-
terion for distinguishing correct from true descriptions.
Think for a moment what might be entailed byra candid,
negative answer on the part of the agent to the question
'Are you X-ing?'-or 'Did you X?E- First, a negative answer
to this question may simply constitute a truthful denial
that the description in question applies to the act under
consideration. This case presents no problem; it is simply
an instance of a false description, it does not accurately
describe what the agent did. In the above case, for ex-
ample, a negative answer on the part of the fireman to the
question 'Were you firing the boiler?‘ accurately reflects
that this description is false.

But what else might a negative answer to the ques-

tion 'Were you (are you) X-ing?‘ indicate? It seems clear

41

that it may indicate simply a lack of awareness, delibera-
tion, heed, or attention on the part of the agent. If our
fireman had told us that he was not swearing, we might
rightly respond, 'You were swearing all right—-even if you
weren't aware of it.‘ In other words, on Rayfield's
analysis it is difficult to see how we are to understand

a locution of the form 'A did X unwittingly.‘ To relegate
such unwitting performances to the realm of 'unbound ac-
tions' is unhelpful. The act still 'belongs'~to the agent
in the sense that he is the author of the act, and that

we hold him responsible for its.commissiona«-Though Ray-
field's analysis does not go quite so far as to entail
that all acts are (by definition) intentional, it does
entail that all acts are performed with heed or attention
on the part of the agent; but even this weaker claim is
false.

In response to these criticisms, Rayfield38 charges
me, among other things, with failing to understand his dis-
tinction between bound and unbound actions-and the corre-
Sponding distinction between true and correct act-
descriptions, as well as failing to grasp the purpose for.

which he draws these distinctions in the first place.

42

First for the distinctions themselves. Rayfield quotes my

claim that

. . . a negative answer to this question

[viz. 'Are you X—ing?‘ or 'Did you X'] may

simply constitute a truthful denial that the

description in question applies to the act

under consideration. This case presents no

problem; it is Simply an instance of a false

description, it does not accurately describe

what the agent did.39
He takes this statement as evidence that I do not compre-
hend his distinction between true and correct act-
descriptions:

Miller apparently takes it that the appro-

priate negativexanswer«to»this question is

both necessary and sufficient-to qualify

the description 'a_Xfed,' as-a false de-

scription of 5's behavior.40
In the first place, I do not believe that-the appropriate
negative response is both necessary and sufficient to dis-
qualify a given description from being true. Furthermore,
and more importantly, I deny that the passage-which Ray-
field quotes commits me to this claim. On the contrary,
I understand quite well that a negative answer to Q4 (i.e.
'Are you X-ing?‘ or 'Did you X?') is not sufficient for
qualifying a putative description as false. In fact, I

was careful in the passage Rayfield cites to use the word

'may'--not 'must'--in the phrase "may simply constitute a

43

truthful denial that . . . ." I realize that a negative
answer to the crucial question needvnot-indicate that the
description under consideration is false (although, on
Rayfield's analysis, it is sufficient to qualify the de-
scription as incorrect). In fact, I explicitly consider
the other alternative in my final paragraph;4-l namely,
that although the given description may be true, a negative
answer may simply indicate a lack of heed or awareness on
the part of the agent. -In short, Rayfieldhs~claim that
"the puzzlement expressed in Miller's last paragraph is
understandable only if it's assumed that appropriate nega-
tive answers to Q4 are both necessary and sufficient to
qualify a given description as false"42 is clearly without
any foundation whatsoever. On the contrary, if this la
assumed then my comments in the final paragraph are indeed
unintelligible.

If Rayfield's first charge is based on a misunder-
standing, his second charge is right on target and serves
nicely to locate the focal point of-our dispute.' With re-
spect to the purpose for which he draws the bound/unbound
and true/correct distinctions in the first place, I quote

Rayfield:

 

44

In certain circumstances, theothings~people
do take on a special importance:' behavior,
which in other circumstances might pass
without notice, becomes subject to praise
or blame. Before the actual bestowal of
praise or blame can be justifiedf-however,
it is necessary that we know that the be-
havior in question qualifies as what I have
called a bound action. Moreover, in such
cases, it is not enough to know that the
description(s) used for the behavior are
true; we must also know that the descrip-
tions match that-or those of the person
whose behavior is in question. Otherwise,
we are confronted with what I S2 take to
be a disturbing paradox: we are apt to
bestow praise or blame for something done
unwittingly or unknowingly.43

In other words, Rayfield's reason for distinguishing-bound
actions to which 'correct' descriptions may be applicable
from unbound actions to which "merely true" but not correct
descriptions apply is to distinguish those items of human
behavior which are amenable to praise or blame from-those
which are not. Rayfield makes this point explicitly later
in his paper:

Generally, the motivation [for making these

distinctions] is provided when we are about

to praise and reward or blame and punish

others or ourselves for behavior--because

. . . we only praise and reward or blame

and punish behavior which qualifies as
(bound) action.44

In response to another critic, A. B. Cody, Rayfield takes

the opportunity to make this same point, this time even

45

more emphatically: “I consider it a necessaryicondition
of ascribing praise or blame to someone's (A's) behavior,
X, that X qualify as A's bound'action."45

This claim Seems to me to be clearly mistaken.
This is not the point to enter into a detailed examination
of the conditions for the ascription of-praise and blame.
However, I would like at this time to briefly discuss my
reasons for rejecting this view.

Let us begin by considering an illustration offered.
by Cody.46 Suppose an outfielder inra game*of baseball
throws out a runner at home base.. According to Rayfield,
while 'Smith threw to home base' is a correct as well as a
true description of what Smith did, 'Smith saved the tie,‘
while true, is nevertheless unlike the former not a correct
description of Smith's action; presumably, Smith would
answer the question 'Are you X-ing?‘ affirmatively in the
former case (X = 'throwing to home base'), but would not
or could not answer affirmatively in the latter case (X =
'saving the tie'). Furthermore, given his statement of the
necessary condition for ascribing praise and blame, it
follows that Rayfield would "find it difficult to believe
that we would praise Smith for saving the tie . . . if, as

is the case by hypothesis, what he did was unwittingly

 

46

done."47 I agree with Rayfield on this; I too would find

it strange to praise someone for something done unwittingly.
Nevertheless, I reject Rayfield's statement of‘the necessary
condition for the ascription of praise and blame. My reason
for this is that in stating the condition-Rayfield seems to
be assuming a symmetry between the conditions for ascribing
praise, on the one hand, and ascribing alama, on the other.
It seems clear to me, on the contrary, that the two-are not
wholly symmetrical, that the conditions for ascribing blame.
are not identical to the conditions for ascribing praise.
Here I am following D'Arcy who claims-that good and bad

acts are asymmetrical with regard to the circumstances of
intentionality and consciousness (i.e. heed,-awareness, or
knowledge on the part of the agent).48 The-absence of
intentionality or consciousness, according to D'Arcy, is

not always an excusing circumstance (with respect to bad

actions), but their absence is always a disentitling cir-

 

cumstance (with respect to good actions). That is to say,
if A is not aware of X-ing while he is X-ing, then'he is
not entitled to any praise which might otherwise be-forth-
coming. If we read 'A is conscious of X—ing while he is

X-ing' in Rayfield's terms as 'A knows that he is-X-ing

47

while he is X-ing,' then I think the'claim concerning the
asymmetry between ascribing praise andiascribing*blame can
be stated quite simply. That is, though we may (sometimes)
blame A for X-ing even though‘he was not conscious of X-ing
(i.e. A would answer our question 'Were you or are you
X-ing?‘ negatively), we nevertheless do not praise a person
for X-ing if he didn't know he was X-ing. So if A X-es
unwittingly, and even if X (or the consequences-of-X) is
felicitous, we nevertheless do not praise A for X-ing; we
simply don't praise a person for doing good accidentally.49
Therefore, with respect to Cody's baseball illustration, it
is not surprising that we would not praise Smith for saving
the tie if what he did was done unwittingly. But from this
it does 222 follow that we would (or could) never=blama A.
for X-ing even though he X-ed unwittingly.50

This is the point at which our disagreement mani-
fests itself most markedly. For if what I have said is
correct then it follows that we may be blamed not aaly for
our (bound) actions which are culpable, though-it does
follow that we can only be praised for (bound) actions.
Specifically, I am suggesting that one may—-not must-~be

held responsible and blamed for at least one species of

48

unbound action, namely, that generated by appropriate nega-
tive answers to the question 'Are you X—ing?’ or 'Did you
X?‘ (which answer, as pointed out earlier, may simply indi-
cate a lack of heed or awareness on the part of the agent).
Finally, I do not think that anything I have argued

for here commits me to the "limited and strange view" that

. . . if anything an agent does qualifies as

an action it might be something tey [he] does

unknowingly, or that everything tey [he] does

knowingly thereby qualifies as an action of

ters [his].51
So much then for Rayfield's analysis. Let us now turn our
attention to a second proposal for selecting the correct

act-description from all the true descriptions applicable

to a given action.

(ii) Intention and Convention

Andrew Oldenquist has discussed two-independent
criteria for identifying the 'correct' (i.e. apprOpriate)
'description of an act.52 He begins by considering the
illustration of a person (let us call him Jones) extending
his arm from his automobile. Any number of descriptions
may be applicable to this single event (or cluster of

events). For example, 'stretching his arm,‘ 'signaling a

49

left turn,‘ and 'pointing to a church' may all be taken to
be compatible with the set of events observed (assuming, of
course, that there is a church within Jones' field of vi-
sion). On what basis, then, are we to-decide which of
these descriptions, if any, correctly describes what Jones
is (was) doing on this occasion? According to Oldenquist,
"since, by hypothesis, which of these three actions Jones
actually performs does not depend on events he produces, it
behooves us to seek elsewhere for criteria that justify the.
application of one action description rather than-another."53
And the two criteria he considers are "the applicability of
social practices or conventions and the intention with
which the action is done."54 What this means is that "given
other conditions, the existence of an intention or an appli-
cable practice logically entails . . . a given action."55
Let us turn then to an examination of these criteria and
attempt to determine whether they fare better than that
suggested on Rayfield's analysis.

To begin with, it is clear that certain types of
actions are logically dependent on the existence and opera-

tion of various social practices and conventions.- Paradigm

cases of such act-types would include such things as

50

signaling turns from one's automobile, making a move in
chess, repaying one's debts, and so on.- It is no less
obvious that an agent's intentions often play a crucial--
though not necessarily always a decisive--role in determin-
ing what act is or has been performed. However, it can
easily be shown that applying the two suggested criteria
for identifying the correct description of a given action
may lead us to inconsistent results. For example, consider
the above illustration. On the one hand, given Jones' ex-
tension of his arm, if he intended to pointrout a church to
his companion, then, presumably, this entails that he was
pointing to a church (the existence of the intention on
Jones' part). On the other hand, given-the existence of
the practice or convention of signalling turns from one's
automobile, and given Jones' extension of his arm, this
would, on the analysis under consideration, entail that
Jones was indeed signalling a turn (the existence-of an
applicable convention). In other words, to state the dif-
ficulty more generally, it may sometimes be the case that,
given a set of physical occurrences (in this case, Jones'
extension of his arm), the (description of the) action

logically entailed by the existence of an intention on the

51

part of the agent may not be the same as the (description
of the) action entailed by the existence of an applicable
practice or convention. In our illustration,-Jones' inten-
tion to point out a church to his companion entailed that
he pointed to a church, while the existence-of the practice
or convention of signalling turns from one's automobile
entailed that Jones was signalling a turn.56 Oldenquist
himself, in all fairness, seems to be sensitive to just
this sort of difficulty:

But the issue is complicated by the fact that

given an appropriate bodily movement, it ap-

pears to be sometimes sufficient and sometimes

merely necessary for an action to take place

that one intend it. Similarly, it appears to

be sometimes sufficient and sometimeswmerely

necessary for an action to take place that a

practice or convention apply to it, and some-

times the convention and the practice will be

jointly sufficient. It will be evident that

which of these alternatives holds depends not

only on the kind of action in question but

also on the circumstances in which the~action

is to be done.57
All this seems correct. It is not possible, for example,
to commit murder unintentionally, but I can nod my head
unintentionally. Given certain circumstances, however,
nodding one's head = bidding at an auction. ~In this case,

the circumstances and the relevant practices and conven-

tions dictate that one has made a bid whether'he intended

52

to or not; he will be held responsible for making the bid
and expected to pay up. So I agree with Oldenquist on this.
All I am claiming is that to admit this "complication" is
tantamount to giving up the claim that either of these two
proposed criteria does, in fact, succeed in establishing
necessary and sufficient conditions for identifying which

of many possible descriptions correctly applies to a given
bit of human behavior. The precise role of intention and
convention in determining the correct description applicable
to a given action will be taken up in detail in Chapter III;
for the present, I have only attempted to show that they are
not necessarily EBB decisive factors for identifying correct

act—descriptions.

4. Description and Purpose

 

In the preceding section we examined two proposed
models for identifying correct descriptions of human actions.
I have argued that both of these models are unable to with-
stand analysis, and consequently fail to do the job required.
During the course of this examination, I considered each

proposal in turn and offered arguments directed specifically

53

against various shortcomings in each. I should now like to
turn to a more general discussion of all such proposals and
try to show that, in addition to whatever specific failings
are unique to any proposed analysis of this type, there is
a further assumption which is implicit in all such attempts;
and furthermore, that this assumption constitutes a fatal
error which renders all such attempts wrong-headed from the
start. I shall take it as obvious that if one attempts to
determine a set of necessary and sufficient conditions for
distinguishing 'correct' from 'true' act-descriptions in
the way that we have been using these terms, that person a
fortiori assumes that such a specification is possible; in
other words, he does not believe that his enterprise is
doomed from the outset. I would like to show why I feel
that this assumption is wrong, and why those who proceed on
the basis of this assumption must find themselves confronted
with an impossible task.

At the risk of being redundant, let me state ex-
plicitly once more the desideratum of our enterprise; The
starting point for our inquiry is the hypothesis that any
given act may be truly described in many different ways.

Among this class of many true descriptions applicable to a

54

given act, however, not all will count as 'correct' de-
scriptions as well of the act in question. Here 'correct'
is being taken to mean, roughly, 'appropriate in a given
context.‘ What we want is some method or procedure for
picking out, as it were, from the class of true descrip-
tions that description (or those descriptions) which is
not only true but correct as well. Now I think that before
one proceeds to attempt to distinguish correct from true
descriptions, it would first be wise to get clear on the
notion of description itself-. Once this concept has been
sufficiently attended to, I think it will become clear why
the sorts of analyses discussed above are misguided in
principle.

It is extremely important at the outset to recog-
nize and emphasize the fact that descriptions are not asked
for and given in vacuo; when one asks for or offers a de-
scription of some event, he wants or offers the description
for some purpose. A scientist, for example, may-be inter-
ested in the occurrence of some natural phenomenon-for any
number of reasons. He may be seeking a confirming instance
for a scientific law or hypothesis, he may wish‘to-be able

eventually to explain the event in question, he may want to

55

base further predictions on its occurrence, andiso on.
Which of these he adopts as his aim or purpose will deter-
mine to a great extent what sort of description of the
phenomenon in question he will give, which features of the
event are particularly relevant and ought to be mentioned.
That is to say, the correct or apprOpriate description in
a given context is, at least partly, a function of the in-
quirer's interests,.aims, or goals. ~Failure to recognize
this rather elementary fact about the notion of‘a descrip-
tion can result in much confusion and argument at cross-
, purpose;--ThiS-feature—of description holds no less true
for that class of events which we single out as human
actions. It is a feature which attends thesconcept of de-
scription itself, and it obtains regardless-of the type of
event (or object) to be described. In other words, al-
though the problem of act-descriptions does involve certain
questions about factors peculiar to human action,-it is
nevertheless in large part simply a species of the problem
of description itself.

Depending on one's interests, the description one
offers of some event (say a human actionr may be more or

less general, expanded to include as many details as he

56

deems relevant to his purposes, and-so on; Theremis no way
of determining a priori what will countias-the (or an) ap-
propriate or correct description; we must first know to
what use(s) the description is to be put. As pointed out
earlier in Section 1, it is a unique-feature of human ac-
tions that we are often interested in their descriptions
for purposes of moral and/or legal assessment. But even
where we have succeeded in specifying our purpose to this
extent, there is still considerable flexibility in what we
might count as an appropriate or correct description of a
given item of behavior.. For instance,-I may be interested
only in the most generalacharacterizatien placing the act
in question in a "moral genus" ('X was an act of injusticefl
'Y was an act of mercy,‘ etc.). Or on the other hand, I
may be interested in locating the act in question in its
relevant "moral species." I may already know, for example,
that A killed B, but what I want to know is whether the
killing in question constituted murder; that is, I want to
know whether A killed B intentionally, with malice afore-
thought, and so on.

From these considerations it should be clear that

the notion of a 'correct' description is highly

57

context-dependent. We cannot begin to answer thetquestion
whether a particular description iscorrect or appropriate
until we have first determined the answer to=a prior ques-
tion, namely, 'Appropriate for what?‘ Do you want a de-
scription of what the agent did which'would-satisfy a
physiologist, or one which the jurors in a court of law
would find informative? And as fortthe physiologists
themselves, the description which satisfies A who is
working on the neural structure may not be helpful at all
to‘B who is investigating muscular control.» It is this
sort of context-dependence of act—descriptions, I contend,

which makes any attempt to specify necessary and sufficient

 

conditions for correct act—descriptions impossible in prin-
ciple. It is not simply that the task of devising a can-
onical form to serve as a quasi—mechanical-method for
picking out correct descriptionsof human actions would

be an extremely long, complicated, and arduous enterprise;
rather, the enterprise itself is in principle-wrong-headed.
The description picked out as correct or appropriate for
one purpose and in one set of circumstances may not be
appropriate given another purpose and a different set of

circumstances.

58

In light of these considerations, are~we~forced to
conclude that nothing can be gained-in-an-attempt to specify
in what way a certain act-description may-be said to be a
'correct' description of a given act? I do not think that
we must be forced to this conclusion. -To-argue that it is
impossible to establish a set of necessary and sufficient
conditions for correct act—descriptions-is not-to establish
that the notion itself is meaningless or hOpelessly obscure.
At the very least we can~beginf~following-Rayfield, by spe-I
cifying one important necessary condition for a correct
'-act-description; namely, the condition that such a descrip-
tion must be papa. That is to say, an act-description is
correct only if it is true. If a putative description of a

given item of human behavior is false, then that description

 

is a fortiori not correct. For instance, in the fireman
illustration in the last section, to describe his act of
extinguishing a fire as 'firing the boiler' is Simply to
offer a false description; that is app what the fireman
was doing, indeed the two are mutually incompatible; This
description has failed to meet the minimally necessary
condition for a correct description. -But of course the

converse entailment does not hold; that a certain

59

description satisfies the condition of being true does not
thereby ensure that that description is also correct. But
this should not be surprising in light of my contention

that no determination of necessary and sufficient condi-

 

tions for correct act-descriptions is possible. On what
grounds, then, are we to proceed in attempting to discover
what la involved in the notion of a correct description?
What I want to suggest is that we analyze the no-
tion of a correct (act) description along the lines of a
defeasible concept, to borrow from.Hart's,terminology. I
have already mentioned one necessary condition which any
given description must satisfy if it is to qualify as a
correct description; that description must at least be
true. Moreover, I have foregone any attempt to specify
further conditions which would, if conjoined with truth,
constitute a set of necessary and sufficient conditions.
Nevertheless I do feel that we can point to various sorts
of defeating conditions which would disqualify certain true
descriptions from being correct as well.
We have been taking the term 'correct' to be roughly
equivalent to 'appropriate.’ Now one important way in which

a given description may fail to be apprOpriate is if it is

60

in some sense misleading. Given this, we may provisionally
characterize a certain description of an itemiof‘human be-
havior as a 'correct' act-description if it is (i) true,
and furthermore (ii) it is not misleading or infelicitous.
Those features which would render a given (true) descrip-
tion infelicitous would be precisely-those sorts of defeat-
ing conditions whose presence would nullify a description's
claim to correctness.

I am well aware that all this-talk about a descrip-‘
tion (or any type of statement, for that matter) being in-
felicitous or misleading commits me to something akin to
Grice's concept of "conversational implication" or to a
notion like "pragmatic implication" often associated with
Moore. At any rate, such talk commits me-to aamanotion of
"implication" distinct from strict logical implication or
entailment as these are understood by the formalist. Nor
am I oblivious to the fact that some philosophers regard
these notions at least with suspicion, if not outright
contempt. The grounds for such suspicion and contempt,
however, are not always so clear. At times it seems as if
those who find themselves opposed to all sorts of non-

logical implication are intent on denying that there is

61

any such thing at all. This seems to me to be~c1early mis-
taken. At other times, however, those who are inimical to
the concept of non-logical implication-simply appear to
bemoan the fact that this concept does not lend itself to
the kind of precise and rigorous-analysis with which they
are so enamored. And this I am prepared to concede. For
non-logical implication, whatever else it involves, is a
relation holding between statements in a natural language;
to attempt to subject this to rigorous analysis byrimposing'
the straitjacket restrictions of an artificial language is
to blur those very features of-English words and sentences
which make the notion of some sort of non-logical'implica-
tion useful to begin with. But to admit this is not to
concede that the concept itself is and must remain hope-
lessly vague and obscure; it is only to admit that an
adequate explication of this concept may not be accomplished
with the facility that some might like. We will not be able
to ask questions like 'What does any sentence of form F
(non-logically) imply?'; rather, we must in each instance
ask 'What is (non-logically) implied by gala English sen-
tence?‘ This does not mean, however, that the results of

our inquiry will be valid only with respect to

62

act-descriptions formulated in English. The emphasis is not
on English as opposed, say, to French or Swahili, but on
natural languages as Opposed to artificial languages. Nor
should my comments be taken to suggest that no general con—
clusions will be forthcoming. Various terms and phrases
which we use to designate certain act-types can be clas-
sified, for instance, with respect to whether or not they
entail the presence of intentionality and consciousness.

If we can succeed in spelling out the-relations between
these and other concepts peculiar to human actions such as
voluntariness, involuntariness, ascription of responsi-
bility, praise and blame, the presence and absence of
certain motives, and so on, we may then find ourselves in

a position to say what is and what is not (non-logically)
implied by an act-description employing one or more of

these terms and phrases. And given this, we would then

be able to determine whether a certain description is or

is not correct. All of this will become much clearer when
we get down to actual analyses of cases. Before proceeding,
however, I feel it is necessary to give at least a rough

Sketch of what I take non-logical implication to be.

63

The idea behind the notion of non—logical implica—
tion is basically this: when I say something, I am com—
mitted not only to the actual content of the sentence I
utter in the sense that I must mean what I say,58 but also
to the relevant implications of what I said in the sense
that I am committed to that which I legitimately give my
hearer(s) to understand by the statement I-utter. A couple
of examples may help to clarify this. Suppose that I in
fact know that Morrill Hall was constructed in l9___; if
I then proceed to tell someone 'I believe that Morrill Hall
was constructed in l9__,' then what I have said is not
strictly false,59 but it is nevertheless in an important
sense misleading or infelicitous. For I have then given
my hearer to understand that I "merely believe" the propo—
sition in question. And this holds in spite of the fact
that belief does not formally entail absence of knowledge;
indeed if, as most philOSOphers agree, knowledge entails
belief, then how gaala belief entail the absence of knowl-
edge? In other words, saying that one believes that p
non-logically--though not formally--implies that one does
pal know that p. To take another case more germane to our

purposes, consider the following illustration.‘ Suppose in

64

a court of law that the accused protests that all he did
was plunge a knife into a motionless object; this descrip-
tion of his act, while true, may nevertheless be misleading
if he fails to mention that the motionless object in ques-
tion was his sleeping wife. It is this sense of misleading
or infelicitous which I wish to exploit in attempting to
distinguish correct from "merely true" act-descriptions.60
To repeat again, what will count as misleading or
infelicitous cannot be determined a priori; it will vary
greatly depending on the circumstances and the needs and
concerns of those interested in the description of a given
act. What is important in one context, say the presence
or absence of intention on the part of the agent, may not
be relevant in another. Certain descriptions may be neu-
tral with respect to relevant aspects of action; in such
cases, whether further specification is required will be
a function of the interests and circumstances. Some de-
scriptions, on the other hand, themselves entail at least
partial specification of various relevant aspects; murder,
for instance, unlike manslaughter, entails the presence of
intention on the part of the agent. Whether and, if so,

the extent to which a given act-description must be

65

expanded or contracted in order to avoid being misleading
will again, of course, depend on the given act, the rele-
vant circumstances attending its execution, the context
in which it takes place, the needs and interests of those
concerned, and so on. With all this in mind, then, let
us now turn to the problem of act-descriptions as it re-

lates to the issue of actions and consequences.

NOTES ON CHAPTER I

l. The Function of Action-Sentences

1A. I. Melden, "Action," The-Philosophical Review,

Vol. LXV, No. 4 (October, 1956), p. 530.

 

2Even if it were necessary that I need to know what
bodily movements took place-in order to know what I was
doing in any given case, it would app follow that I could
know this-only by observing my own movements; that is, if
we allow for the possibility of "non-observational«knowl-
edge," to use Miss Anscombefls phrase.

3John W. Yolton, "Ascriptions, Descriptions, and
Action Sentences," Ethics, Vol. 67, No. 4 (July, 1957),
p. 307.

4Melden, p. 541 (italics mine).

5John W. Yolton, "Act and Circumstance;" Journal of
Philospphy, Vol. LIX, No. 13 (June 21, 1962), p. 338.

6

Yolton, "Ascriptions, Descriptions, and Action
Sentences," p. 307.

 

 

7Stuart Hampshire, Thought and Action (New York:
The Viking Press, 1959), p. 90.

8P. T. Geach, "Ascriptivism;" The Philosophical
Review, Vol. LXIX, No. 2 (April, 1960). p. 221:° This paper
represents a general critique of the sort of position held
by Hart and Melden, though Geach himself mentions no one in
particular (other than a general reference to "some Oxford
philosophers"). George Pitcher also criticizes this sort
of position in his article, "Hart on Action and

 

 

66

67

Responsibility," in the same issue of The Philosophical
Review, pp. 226-235.

 

9Yolton, "Ascriptions, Descriptions, and Action
Sentences," p. 310.

10There are, of course, some limits to this flexi-
bility. I could hardly use a locution such as 'How many
people are inside?’ to issue a command, for example. Even
here, however, I suspect that the flexibility is greater
than is commonly supposed. A rhetorical question, for ex-
ample, though it shares the interrogative form of other
questions, is not really a question at all in the way it
functions.
11In a similar vein Anscombe distinguishes the point
of a sentence in the language from the speaker's purpose in
uttering the sentence (Intention, p. 3). Thus, a sentence
may have more than one "point" in the language, even though
the speaker on a given occasion may be exploiting only one
of these "points" in uttering or using the sentence. This
is for all practical purposes equivalent to my distinction
between the "functions" of a sentence and the illocutionary
act performed by a speaker in uttering the sentence on a
given occasion. (Of course I am not forgetting that, to
paraphrase Strawson, sentences don't do anything, rather it
is people who use sentences to do things.)

len the face of it, this may seem to contradict my
earlier claim that moral responsibility does not entail
causal responsibility. But this is not the case. My
earlier remarks concerned general claims about the ascrip-
tion of responsibility in its various senses, not the func-
tion of sentences of a certain form. The present discus-
sion is restricted to the jobs performed by action sentences
of the form 'A did X.‘ Thus, in our previous example,
though A may be morally responsible for his wife's suicide,
the sentence 'A's wife committed suicide' itself ascribes
no responsibility to A. Of course, A's responsibility may
be captured in a redescription of his action in a sentence
of the form 'A did X' or 'A X-ed' (e.g. 'A drove his wife
to suicide'), in which case the present analysis applies;
in the ascription of moral responsibility to A he is a
fortiori identified as the author of the act in question
(viz. driving his wife to suicide).

 

68

13Joel Feinberg, "Action-and~Responsibility," in
Philosophy of Action, ed. Alan R; White (Oxford University
Press, 1968b p. 110.

 

2. One Act--Many Descriptions

l4Eric D'Arcy, Human Acts: An Essay in their Moral
Evaluation (Oxford: Clarendon Press, 1963), p. 10.

15

 

John L. Austin,"A Plea for Excuses," p. 148.

16Keith Donnellan, "Knowing What I am Doing," Journal
of Philosophy, Vol. LX, No. 14 (July 4, 1963), p. 401.

17Brian O'Shaughnessy, "Observation and the Will,"
Journal of Philosophy, Vol. LX, No. 14 (July 4, 1963),
p. 375.

l8Stuart Hampshire, Thought and Action (New York:
The Viking Press, 1959), p. 169.

19See also, for example, G. E. M. Anscombe, Intention
(Oxford: Basil Blackwell, 1958), p. 11 ("Since a single
action can have many different descriptions . . . "); and
Charles Taylor, The Explanation of Behavior (London:
Routledge & Kegan Paul, 1964), p. 28, n.

20Hampshire, p. 169.

21D'Arcy, p. 12.

 

 

 

22Nicholas Rescher, "Aspects of Action," in The
Logic of Decision and Action (Pittsburgh: University of
Pittsburgh Press, 1967), p. 215.

23It is this sort of possibility which leads Kenny
to observe that "If we cast our net widely enough, we can
make 'Brutus killed Caesar' into a sentence which describes,
with a certain lack of specification, the whole history of
the world." Action, Emotion, and Will (London: ~Routledge
& Kegan Paul, 1963), p. 160. And it is considerations like

 

69

this which lead him to argue against the temptation to
treat actions simply as a queer species of polyadic rela-
tion (Ch. 7, "Actions and Relations").

24D'Arcy, p. 1.

25This sort of case will be discussed thoroughly in
Chapter II, Section 2.

26At least not in 1943. Unfortunately, modern
weapons technology appears to have made it possible to
commit genocide at "one fell swoop."

27Donald Davidson, "Actions, Reasons, and Causes,"
Journal of Philospphy, Vol. LX, No. 23 (November 7, 1963),
p. 686.

 

28Arthur B. Cody, "Can a Single Action Have Many
Different Descriptions?," Inquiry, Vol. 10, No. 2 (Summer,
1967). p. 165.

3. Correct vs. 'Merely True'
Act Descriptions

29With a few minor stylistic variations, this portion

of the present chapter (PP. 33-41) has been published under
the title "Correct vs. 'Merely True' Act—Descriptions" in
Inquiry, Vol. 17, No. 4 (Winter, 1974), pp. 457-460.
laguiry is published by Universitetsforlaget, Oslo, Norway.

30David Rayfield, "On Describing Actions," Inquiry,
Vol. 13, Nos. 1-2 (Summer, 1970), pp. 90-99.

31Rayfie1d, p. 93.

32Rayfield, p. 93.

33Rayfield, pp. 92-93.

34Rayfield, p. 94.

35Rayfield, p. 95.

70

36Rayfield, p. 93.

37Rayfield, p. 91; for a more detailed discussion
see Rayfield's "Action," Nous, Vol. II, No. 2 (May, 1968),
pp. 131-145, esp. pp. 136-145. Also see Rayfield's more
recent Action: An Analysis of the Concept (The Hague:
Martinus Nijhoff, 1972), particularly chapters I, II, and
V, as well as my review of same in The Michigan Academician,
Vol. VII, No. 1 (Summer, 1974), pp. 128-30.

38David Rayfield, "On Miller's Paradoxes and Circlesfl'
Inquiry, Vol. 17, No. 4 (Winter, 1974), pp. 461-464.

39

p. 40.
4oRayfield, "On Miller's Paradoxes-and Circles,"
p. 462.
41pp. 40-41.
42Rayfield, "On Miller's Paradoxes and Circles,"
p. 462.
43

Rayfield, "On Miller's Paradoxes and Circles,"
p. 461. Even if this result were disturbing, I fail to
see how it constitutes a paradox.

44Rayfield, "On Miller's Paradoxes and Circles,"
p. 463.

45David Rayfield, "Cody's Categories," Inquiry,
Vol. 16, No. 4 (Winter, 1973), p. 425.

46Arthur B. Cody, "Is 'Human Action' a Category?,"
Inquiry, Vol. 14, No. 4 (Winter, 1971), pp. 404-405.

47Rayfield, "Cody's Categories," p. 425.

48D'Arcy, Human Acts: An Essay in Their Moral
Evaluation, Ch. 3.

 

71

49We might praise a child, for example, for doing

something we regard as commendable, eveniif the child were
not aware of doing that for which he or she is commended.
But a case like this seems to be more akin to encouragement
than to commending per se; we point out to the child those
features of the action which make it praiseworthy, and then
"commend" or reward the child—-in words or otherwise--in
order to encourage the same sort of behavior in similar
circumstances in the future. With respect to cases such
as these, it seems useful to distinguish-calling an act
good (or bad) and praising (or blaming) the agent for per-
forming the act in question.

50Rayfield seems to be guilty of another sort of
asymmetry as well. In his analysis, Rayfield implies that
either of two questions, namely 'apa you Xing?‘ and 'Qla
you X?‘ will do equally well for distinguishing bound from
unbound actions (and true from correct descriptions). But
his baseball illustration highlights particularly well the
fact that the present tense and past tense uses of the
question are app on a par. If we construe ‘saving the tie'
as an achievement verb (to use Ryle's terminology), then of
course Smith would not be in a position to answer the ques-
tion 'apa you saving the tie?‘ affirmatively--or negatively.
Being an achievement verb, he could not answer the question
at all until such time as the feat is either accomplished or
he fails. But I would think that there would be no impro-
priety in Smith's answering the question, 'pla you save the
tie?‘ affirmatively, assuming of course that the throw was
successful. Perhaps Rayfield would object that 'saving the
tie' is not something that anyone could decide to do; That
is, although Smith could decide to pay to save the tie,
'saving the tie' is not itself something he could decide
to do. But if that's the case, then it seems that the same
considerations ought to apply to 'throwing to home base' as
well (if we construe this as an achievement verb too).
Strictly speaking, Smith is entitled to assert only that he
is trying to throw the ball to home base; after all, the
ball may fall short of second. Again, of course, once the
feat is accomplished, Smith could rightly answer 'Yes' to
my question, 'Qla you throw the ball to home base?‘

On the other hand, it doesn't seem to matter which

of the two questions we ask the agent with respect to task
verbs like 'running' or 'throwing a ball,’ for example.

 

72

Here it seems simply a function of the-time at which we put
the question to the agent; that is, whether we ask him the
question while he is engaged in the activity (present tense
use), or after the doing has been completed (past tense useL
Nevertheless, it does not seem that-the'present and past
tenses are on a par generally, once the distinction between
task verbs and achievement verbs is taken into account. (It
is interesting that Rayfield puts only the present tense use
of the question to Smith in the baseball illustration, that
is, 'éEE you saving the tie?'). -At any rate, even if I am
right about this, it is not as important as the asymmetry
between the conditions for ascribing praise and blame, and

I don't wish to push the matter.

51

 

Rayfield, "Cody's Categories," p. 427.

52Andrew Oldenquist, "Choosing,)Deciding, and Doing,"
The EngyclOpedia of Philosophy, ed. Paul-Edwards (New York:
Macmillan, 1967), Vol. 2, pp. 96-104.‘ I’do not wish to sug-
gest that Oldenquist necessarily endorses this View; he
simply discusses it as a possible solution; The initial
plausibility of his proposal, I feel, merits at least a
brief discussion.'

53Oldenquist, p. 102.

 

54Oldenquist, p. 102.

55Oldenquist, p. 102.

56This problem is generated by treating 'or' in the
phrase "the existence of an intention pp applicable prac-
tice" in the inclusive sense of 'or'; but it is clear from
the context of the passages under consideration that this
is the sense which Oldenquist intends.

57Oldenquist, pp. 102-103.

73
4. Descriptioniand-Purpose

58Not in the sense that I must be committed to the
truth of what I say (otherwise, we would have to exclude
such speech-acts as lying, telling~a tale, and so on); it
is simply that the meaning of a sentence uttered by a
speaker is independent of that speaker's intentions. In
this sense, one can intend to say something-and neverthe-
less fail to say it; and one can also manage to say some-
thing which he never intended to say.

59Here I am assuming that if one knows that p then
one also believes that p.

60Henceforth I shall drop the cumbersome phrase
'non-logical implication' in favor of the word limply'
(and its various forms) flanked.by scare quotes to indicate
that this is what is intended.

CHAPTER II

ACTS AND CONSEQUENCES

It is clear that the-things we do, the actions we
perform, are often attended by various consequences which
are often foreseen and intended but-occasionally unforeseen
or unintended-by the agent~himself.e I stand up and leave
the room, and as a result the.quest speaker at the collo-
quium is offended. I swerve to avoid an oncoming automo-
bile, and unfortunately a pedestrian is struck and killed.
I marry Ms. A and Ms. B is jilted. In these and similar
cases, we often say that my X—ing resulted in Y or, alter-
natively, that Y was a consequence of my act X. It is no
less obvious that one often accomplishes something by doing
something else; stated simply, we often employ various
means for achieving ends which we, for one reason or
another, find desirable or worthwhile. For example, I
close the door by exerting pressure on it with my hand.

Or I discharge my firearm in an attempt to thwart my at-

tacker. In such cases, we commonly speak of an agent doing

74

75

Y by doing X. And in some sense the term‘or phrase which
is substituted for 'Y' in such cases can be*construed as a
consequence of some action on the part of the agent. For
instance, the closing of the door-and the thwarting of my
assailant in the above illustrations are in some sense
results or consequences of my exerting pressure on the door
and discharging my firearm, respectively.- In this chapter
I would like to examine the relation between act and con-
sequence, and in particular attempt to determine the hear-

ing of this relation on the problem of act-descriptions.

1- ARedespribing.Acts in Terms
oftConsequences

 

Frequently when we have an episode consisting of a
number of events which are divisible into an act and the
consequence(s) of that act, we find it convenient to re-
describe the episode in question solely in terms of the
phrase or term used to denote the consequence(s) of the act.
Consider the following sequence: I take aim at Smith with
my revolver, pull the trigger, and kill or-wound him. Al-

though we may quite properly say that 'I shot Smith and

76

thereby killed/wounded him,’ we may just as legitimately
say simply that 'I killed/wounded Smith.‘ Indeed, there
are cases in which this sort of redescription-is not only
harmless and natural, but is expressly called for. For
instance, in our example where I close the door by exerting
pressure on it with my hand,-although«we~may for some spe-
cial reason find it appropriate to describe this episode as
'I exerted pressure on the door with my hand and thereby
closed it,‘1 this description would-under most circumstances-
sound quite stilted and queer; we find ourselves much more
inclined to say simply that 'I closed the door,‘ and this
without any loss. Cases like the two just considered are
numerous; Eric D'Arcy provides us with further illustra-
tions:

For instance, if A stabbed B, and thus killed
him, we may say simply that A killed B. The
act of flicking a switch may produce the con-
sequences that (1) contact is made between
two points, so that (2) current flows from
outside the room, through these points and
the wire in the room, into the globe and
through its filament, and thus (3) the globe
is illuminated; but this may also be described
simply as, 'putting on the light', or 'light-
ing the lamp'. If a person sings a song, and
thus entertains a group of people, we may say
either that he is singing a song, or enter—
taining them; and if as a result of the
latter he raises money for the Red Cross,

we may say that he is raising money for the

77

Red Cross; and if as‘a result of that he

boosts morale, or helps the war effort, we

may also say that he is boosting morale, or

helping the war effort.2
All of these cases are such that we have no hesitation in
redescribing the act in question in terms of one or more
of its consequences. In fact, cases of this sort are so
common and frequent that one may be tempted to conclude
that such redescription is always at least possible in
principle, even if not expressly called for.- It is pre-

cisely this temptation which I would now like to examine

in detail.

(i) Feinberg's "Accordion
Effect" Thesis

 

 

We are agreed, then, that in many cases an action
may be correctly redescribed in terms of its consequences.
We must now ask whether this is possible in every case;
that is, may we always correctly redescribe an act in terms
of the consequence(s) of that act? Some philOSOphers have,
in fact, claimed that such redescription is, in principle,
always open to us. Joel Feinberg, for instance, claims
that this is a function of what he refers to as the "ac-

cordion effect" feature of action. The notion behind this

78

is that "it is a well-known featurevof'ourflanguage;*whereby
a man's action can be described as narrowly orwas'broadly

as we please . . . because an act, like a folding musical
instrument, can be squeezed down to a minimum or else
stretched out." "We can, if we wish, puff out an action to
include an effect, and more often than not our language
obliges us by providing a relatively complex action word

for the purpose."3 Strictly speaking, as Ruth-Macklin

points out, "it is the descrlptions of actions-which can

 

be whittled down or.expanded.in.thissmanner"4-—not the
actions themselves. In a discussion of Feinberg's position,
John Atwell succinctly formulates the "accordion effect"
thesis as it applies to the issue of acts and-consequences
as follows: "If A did X and thereby caused-Y, then (we may
say that) A did Y."5 Feinberg himself concludes that "what—
ever difference exists between the two forms of expression
[between 'A caused Y' (by doing X) and 'A did Y' (by doing
X)] is merely a matter of rhetorical emphasis or grammatical
convenience."6 Having granted that this claim holds true
for a whole host of cases, let us now consider whether this
thesis will do as a qeneral claim applicable to all act-

descriptions.

79

First a couple of observations on strategy. Fein-
berg's thesis purports to be a general claim concerning acts
and their descriptions; as it stands it has the form of a
universal statement. Consequently, citing just one legiti-
mate counter-example would be sufficient for—refuting the
thesis as it stands. But the destructive task is relatively
easy; the job of constructive analysis comes much harder.
The negative criticism by itself, though valuable, con-
tributes little to the solution of the problem of formulat-
ing criteria to govern the redescription of acts in terms
of their consequences. With this in mind, my aim in the
subsequent discussion is twofold. First I will examine
Feinberg's thesis on its own merits, and show why I think
it is false as a general account which purports to hold
across the board for all act-descriptions. Secondly, the
kinds of counter-examples adduced to refute the thesis will,
I hope, provide a basis for constructing general principles
or guidelines for distinguishing cases in which a rede-
scription of an act in terms of its consequence(s) is
appropriate, legitimate, or 'correct' from those cases in
which such redescription must be ruled out as impermis-

sible.

80

If we recall our discussion of-the notion of a
'correct' description in the last chapter, it should be
clear that a given description of an action (or of any
event, for that matter) may fail to be correct in one of
two ways. In the first place, since truth is a minimally
necessary condition for correctness, it follows that if a

given description is false, it is a fortiori not correct.

 

Let us refer to this as the strong sense of-‘incorrect.’
Secondly, a given act-description,.though true, may‘be in
some important respect misleading. Given that a particular
description is true, a further condition for correctness is
the absence of defeating conditions. That is, for a-given
description which is true to count as correct as well, it
must not "imply" or give my hearer(s) to understand or
infer something else which is false, misleading, or unwar-
ranted. It follows that the presence of one or more de-
feating conditions (or "infelicitous implications") pro-
vides a second way in which a description may fail to be
correct. Let us refer to this as the weak sense of 'in-
correct.’ I suppose we could, if we wished, regard false-
hood as the paradigm defeating condition, in which case we

could classify the first sort of incorrectness as a species

81

of the second; in that case, falsehood would be the limit-
ing case of infelicity or incorrectness. But for our pur—
poses it will be much more helpful and illuminating to keep
these two senses of 'incorrect' distinct. Accordingly,
what I now wish to Show is that Feinberg's "accordion ef-
fect" thesis, if taken literally, fails in two ways corre-
sponding to the two ways in which descriptions may fail to
be correct. First, the redescriptions of actions in terms
of their consequences may sometimes result in an act-
description which is incorrect in the strong sense, that
is, act-descriptions which turn out upon examination to be
falaa. Secondly, some such redescriptions, while true,
may nevertheless fail to be correct in the weak sense in-
sofar as they are infelicitous.

The first set of counter-examples to be discussed
in this and the following section are designed to illus-
trate the fact that sometimes a redescription of an act in
terms of the consequence(s) of that act is galaa and hence
incorrect in the strong sense. Let us begin by considering
the following sorts of counter-examples. First, there are
cases in which an agent A cannot be said to have done Y,

even though Y may naturally be construed as a consequence

82

of his act X; these will be cases in which A simply is not
in the position or does not have the authority to do Y under
the circumstances which obtain. Atwell suggests several
cases of this type: We may, he says,

truly assert 'Jenkins caused Murphy to be pro—
moted by informing the corporation head of
Murphy's superior performance record', but we
will deny the action sentence 'Jenkins promoted
Murphy' for the obvious reason that he is not
in a position, i.e. does not have the authority,
to make promotions. And we may affirm 'Baker
bribed certain key municipal officials and
thereby caused the proposal to be approved',
but we will not say 'Baker approved the pro-
posal' because, as above, he is not in a posi—
tion to do so. No doubt he approved of the
proposal and of its being approved, but he

did not approve it. And although a mobster

may have caused a certain defendant to be ac-
quitted by threatening the jurors, he did not
acquit the defendant (especially if he himself
was the defendant).

Notice that the elements in these illustrations square with
the schematic form of the protasis of the "accordion ef-
fect" thesis: l£_a (Jenkins/Baker/the mobster) ala_X (in-
formed the corporation head of Murphy's superior performance
record/bribed certain key municipal officials/threatened

the jurors) and thereby caused Y (Murphy's promotion/ap-

 

proval of the proposal/acquittal of the defendant). Never-

theless, we would not--indeed could not—-conc1ude with

 

Feinberg that we may then say that 'A did Y' (i.e. that

83

'Jenkins promoted Murphy,‘ that 'Baker~approved~the'pro-
posal,‘ or that 'the mobster acquitted-the-defendant').
Not only are these sentences incorrect in the sense of
being misleading; they are simply falaa,

Consider another type of counter-example also sug-
gested by Atwell. There are often cases in which one agent
A, gets something done or accomplished via an intermediary,
B. "In all such cases," Atwell observes, "A does something
X and thereby causes Y to happen, and he causes Y-to happen
by getting B to do Y."8 Suppose, for example, that Jones
gets Smith-to break a window by threatening-him-with phys-
ical harm. In this case, although the-breaking«of the
window is in some sense a result or consequence of Jones's
act of threatening, nevertheless the ensuing breaking of
the window is not anything gaaaa has done; 'Jones-broke
the window' is false. And this holds in spite-of the fact

that Jones may be said to be in some sense responsible for

 

the breaking of the window.9

Finally, consider an illustration employed in
Chapter I. Suppose I, through my acts of cruelty,-negli-
gence, and so on, cause my wife to commit suicide. Again,

although I (A) mistreated and neglected my wife-(X), and

84

thereby caused her to commit-suicide (Y), nevertheless we
must deny the claim that 'I committed my wife's suicide.‘
This case, perhaps even more so thantthe~former examples,
dramatically highlights the impossibility of redescribing
certain acts in terms of the consequences of those acts.
For here we encounter a strict logical-impossibility; the
sentence schema 'A committed B's suicide' (where A and B
are distinct persons) is analytically-false or contradic-
tory. This case has certain affinities with the first set
of counter-examples, such as the illustration concerning

' the mobster and-the defendant's acquittal. If we assume
that the mobster is, in fact, the defendant in the pro-
ceedings, then it is impossible for him to acquit himself,
since it is (logically) impossible for a defendant to
acquit himself. But there is an important difference be-
tween the two. Whereas the mobster qaa defendant cannot
acquit himself, it is quite possible that the mobster aaala,
in other circumstances, be a juror at a trial and thus be
in a position to acquit some defendant. So it is not the
mobster qaa_mobster who cannot acquit the defendant, but the
mobster qaa defendant or qaa non-juror.10 But in our sui-

cide example, it is not simply that given a set of

85

circumstances C which obtain at time t that A'cannot commit
B's suicide; on the contrary, there is no conceivable set
of circumstances in which A ever could commit BLs suicide.
In other words, 'acquitting a defendant' is something a
person pap do (if he is in the appropriate-position or does
have the authority), whereas 'committing~another*s suicide'
is not.

All the cases.considered thus far have-at least
this-much in-common:- these counter-examples are such that
to say 'A did Y' where Y is a result or consequence of some
act X which A performed is to say something false; in none
of these cases can we truly assert that 'A~did Y.J- As such
they show the "accordion effect" thesis to be false as a
general claim concerning actions and action sentences. But
to stop here would be to stOp short; as remarked earlier,
refutation by counter-example, while valuable, contributes
little positive to the solution of the problem of determin-
ing just when and why such redescriptions are incorrect.
Let us begin by asking how or why our counter-examples do,
in fact, tell against Feinberg's "accordion effect" thesis.
Do they have anything in common which might serve as a clue

to what sorts of considerations render this thesis false?

86

We have already made a start in this-direction; for we have
not only adduced ea number of unrelated counter-examples,
but these have been classified under three types. First we
considered cases in which we cannot say that"A did Y' even
though Y was a consequence of his act X because A was not
in a position or did not have the authority to do Y.
Secondly, there were cases where A cannot be said to have
done Y which was a consequence of his act X because of the
agency of some intermediary, B. And thirdly, we considered‘
a case in which A cannot be said to have done Y even though
Y was a consequence of A's act X because 'doing Y"(in our
example, 'committing another's suicide') is something which

A is logically precluded from performing. Let us now at-

 

tempt to determine whether there are any relevant similar—

ities shared by these three types of counter-examples.

(ii) Macklin's "Intervening
Agent" Principle

 

 

One important feature to notice about-all-the sorts
of counter-examples considered so far is that in each case
there has been more than one person or agent involved. In

our first case, even though Murphy's promotion (Y) may be

87

properly regarded as a consequence of Jenkins' laudatory
recommendation (X), Jenkins himself was in no position to

promote Murphy; if Murphy is promoted, someone else other

 

than Jenkins had to do the promoting, someone who, unlike
Jenkins, la in the position or does have the authority to
dole out promotions. In the second case, Jones's threat

was the cause of someone else (namely Smith) breaking the

 

window. And in the final case, my acts of cruelty and

negligence caused-someone else (my wife) to commit suicide.

 

And in all three cases, we have the same kind of “causal
progression": A's original action X (Jenkins' recommenda-
tion/Jones' threat/my acts of cruelty and negligence)
caused someone else, B (Murphy's superior/Smith/my-wife)
to do something else Y (promote Murphy/break-the-window/
commit suicide). The presence of a second agent (B) in
all these sorts of cases provides a possible clue to why
redescription of A's act X in terms of its consequence(s)

Y is to be ruled out as false and a fortiori not correct

 

in such cases.
With respect to cases such as those under consider-
ation, Ruth Macklin has proposed a logical condition for

the redescription of an act in terms of the consequence(s)

88

of that act. She formulates a negative test for such de—
scriptions which she refers to as the "intervening agent"
principle. Stated positively, it specifies a logically
necessary condition for any action of any agent A to be
redescribed in terms of any of its consequences. The
principle is stated as follows:

A description of the consequence of A's

action must not logically precludevits being

a description of theeaction.ll
Let us apply this test to our cases. In the first illus-
tration, although Jenkins was instrumental in the~promotion
of Murphy, and was thus in some sense the "cause" (or one
of the “causes") of his being promoted, we cannot rede-
scribe Jenkins' act (of recommending) in terms of its con-
sequence (Murphy's promotion) by saying that 'Jenkins pro-
moted Murphy' because this (viz. the promoting of Murphy)
is not a description of an action Jenkins (qaa subordinate)
performed or could perform; only Murphy's superior can
promote him. In the second case, though Jones may have
caused Smith to break the window, breaking the window is
nevertheless something §mllh--not Jones--did. -And finally,
though I was, through my acts of cruelty and negligence,

the cause of my wife's death, we cannot say that I

89

committed my wife's suicide, for this is an action which I
am logically precluded from-performing.l2 In each of these
cases, observes Macklin, "the action of another agent act—
ing intentionally renders the second action performed the
action of the second agent. Even if the first action can
be said to have caused the second . . .-it does not follow
that the agent in the first action is the agent in the
second."13

We have now succeeded, I think, in-specifying one
important negative condition for redescribing acts in terms
of their consequences. We have seen that any such rede-
scription which-violates the "intervening agent“ principle
results in an-act-description which is incorrect in the
strong sense, that is, a description which is, in fact,

false. We now proceed to the second species of incorrect-

ness.

(iii) True But
Incorrect Descriptions

 

 

We turn our attention now to considerations of a
different sort. In the previous sections, we were con-

cerned with identifying and classifying some of the ways

90

in which certain types of redescriptions of acts-in terms
of consequences failed to be correct because they were
false; those were cases which failed-to-satisfy the mini-
mally necessary condition for a correct description. We
will now address ourselves to the class of redescriptions
which fail to be correct-—not because they are false--but
because they are in some important sense "infelicitous"

or incorrect in the weak sense; that is, while such de-
scriptions (or redescriptions) satisfy the minimally neces-
sary condition of being true, they nevertheless fail to be
correct for other reasons. To determine what these reasons
are will be one of our principal.tasks.

Once again I will employ the same strategy used in
sections (i)-(ii). I will begin by-taking-Feinber's "ac-
cordion effect" thesis as my point of departure.* I will
then proceed to another refutation of the thesis indepen-
dent of that offered above, again by adducing a number of
plausible counter-examples. Finally, I shall attempt to
analyze and classify the types of counter-examples with a
view to specifying more general principles or-guidelines
for distinguishing those redescriptions of acts in-terms of

consequences which are correct from those which are not.

91

Let us begin our discussion-byrconsidering an
interesting illustration suggested by Prof. J. J. C. Smart
during the racial troubles in Arkansas in 1956.14 We are
to imagine a hypothetical case in which the white people
(presumably in a southern town) believe that a crime has
been committed by some Negro.- However, since they are not
sure which Negro committed the alleged crime, the white mob
decides to lynch five Negroes chosen at random from the
community in order to "teach them a lesson," The local
sheriff is apprised of the group's intentions and, in order
to avoid this catastrophe, he decides to.arrest some Negro,
frame a case against him, have him convicted, sentenced to
death, and executed; the sheriff, presumably, is Opting for
the lesser of two evils. Now there are two distinct ques-
tions we might ask concerning this case. First there is
the ethical issue; we may ask whether the sheriff was mor-
ally justified in so acting. Secondly, we may ask what is
the correct description of the sheriff's act(s). It is
only with the latter question that we are presently con-
cerned.

With respect to this hypothetical case, D'Arcy asks

us to suppose that the sheriff himself acted as executioner.

92

He then considers a number‘of possible descriptions of the

sheriff's act:

1. He tensed his forefinger.

2. He pressed a piece of metal.

3. He released a spring.

4. He pulled the trigger of a gun.
5. He fired a gun.

6. He fired a bullet.

7.. He shot a bullet at a man.

8. He shot a bullet towards a man.
9. He shot a man.
10. He killed a man.
11. He committedtjudicial-murder.

12. He saved four lives.

Now the first thing to notice—about'this list of-descrip-
tions, observes D'Arcy, is that none of them can be re-
jected as simply false. Assuming that each of these is
true, let us focus for a moment on the descriptions toward
the end of the catalogue. For instance, suppose we de-
scribe the sheriff's act as 'committing judicial murder'

(11) and admit that as a consequence of this act he thereby

93

caused four lives to be saved (12). Now let us ask whether
we are prepared to accept the result of applying Feinberg's
"accordion effect" thesis to the case at hand. That is,
given that A (the sheriff) did X (committed-judicial murder)
and thereby caused Y (the saving of four lives), are we
then prepared to assert simply that 'The sheriff saved four
lives'? For my part, I am not prepared to accept this
assertion; even though the sentence is true-(after all,

the sheriff ala save four lives), it nevertheless seems

to me that this description of the sheriff's act-~without
any expansion or qualification-—is in an important sense
misleading.

Let me make it clear at this point that throughout
my discussion of the misleading nature of certain act-
descriptions, I am assuming a background of "ignorance" on
the part of the audience. I want to rule out consideration
of cases in which a particular misleading description, such
as 'The sheriff saved four lives' in the Smart illustration,
EBElS be appropriate or to the point lg those asking for
and giving the description were already fully-apprised of
all the relevant circumstances surrounding the incident.

One can imagine a discussion between the sheriff and his

94

colleagues in which the context may permit one's saying
that 'He saved four lives,‘ particularly if the sheriff's
colleagues acted as his accomplices (although in a case
like that, what would be the point of asking for or giving
a description in the first place?). In other words, given
any act-description, one could, I suppose, always "cook up"
some context--however bizarre-~in which that description
would be to the point. Rather, what I have in mind are
cases in which a person does know of the relevant circum—
stances surrounding some act on his own or another's part
offering a description of that action to a third party who
is ignorant of these circumstances. If A and-B were both
privy to the sheriff's activity, then A's saying that 'The
sheriff saved four lives' would not mislead B, since the
information left unstated in this description is, neverthe-
less, known to both A and B. But if A were to tell C, a
resident of a neighboring town who has heard nothing about
the incident, simply that 'The sheriff saved four lives,'
than this description would be misleading in the sense in
which I am interested.

Before proceeding to discuss the reasons I feel

that the above description is misleading and thus incorrect,

95

I would like to take the Opportunity at this-point to fore-
stall a possible complaint. Lest anyone think that I have
been and am being unfair to Feinberg by misconstruing his
claims concerning the redescriptions of acts in terms of
their consequences, I refer to an illustration suggested by
Feinberg himself.

we can say that Jones Opened the door and
thereby caused Smith (who was inside) to be
startled, in this way treatinngonesls act
as the cause of a subsequent effect; or we
can say (simply) 'Jones startled Smith' (by
opening the door), and thus incorporate the
consequence into the complex action. If
Smith suffered a heart attack and died, we
can say that Jones's Opening the door caused
his death, or that Jones's startling him
caused his death, or.simply that Jones killed
hlg (by doing these things).16

 

Now this passage is to a certain extent ambiguous. In the
last case Feinberg considers, where we may say "simply that
Jones killed him" (viz. Smith), he does not make it clear
whether the parenthetical comment, ”by doing these things,’
is or is not to be included as part of the act-description.l7
But I think there is evidence for believing that Feinberg
does not intend the parenthetical elements to be included.
iFor look how he treats the previous example of Jones's
startling Smith by Opening the door; in that case, he does

not include the phrase 'by Opening the door' within the

96

act-description; rather, he says that we mayrdescribe this
act simply as 'Jones startledISmith:'~ By parity of reason—
ing, then, Feinberg it seems would be'forced to conclude in
the heart attack case that we may simply say"Jone ‘killed
Smith.‘ There is further evidence to support this inter-
pretation in the next paragraph where Feinberg claims that
"instead of saying Smith did A (a relatively simple act)
and thereby caused X in Y, we might say something of the
form 'Smith X-ed Y'; instead of 'Smith opened the door
causing Jones to be startled,‘ 'Smith startled‘Jones.”l8
Again by parity of reasoning, we must conclude that instead
of 'Smith startled Jones causing him to die,‘ we might say
simply 'Smith killed Jones.'19 In summary, then, it seems
that we are to take the "accordion effect" thesis quite
literally; whenever it is the case that A did X and thereby
caused Y, we may say simply that 'A did Y,‘ regardless of
what acts or act-types are substituted for X and Y.

For my own part, however, I am no more prepared to
baldly assert that 'Smith killed Jones' in Feinberg‘s heart
attack case, than I was prepared to assert that 'The sheriff

saved four lives' in Smart's illustration. *This is not,

however, intended merely as an autobiographical-comment;

97

if it were, it would be of no-phiIOSOphical consequence.
Rather, my claim is that most of us would be—-or at least
ought to be--reluctant to say-these things. Our task now
is to explain this reluctance, to advance reasons which
will account for the kind of infelicity which I believe is
engendered in cases like these. We turn now to an exami-

nation Of possible explanations.

(iv).D'Argy's
Elidability Criterion

 

I mentioned earlier that we would have occasion to
discuss D'Arcy's three theses concerning act-descriptions
in detail during the course of our inquiry. We have al-
ready discussed his first thesis which, it will be recalled,
constituted an alternative formulation of the 'one act--
many descriptions' thesis:

Thesis One: there is not necessarily one,

and only one, correct description of a
glven act.26

 

 

 

With respect to any given act, an infinite number Of true
descriptions may be forthcoming (notice that D'Arcy's use
of 'correct' corresponds to my use of 'true'). Within the

class of true descriptions, however, one or more may or may

98

not count as correct as well, depending on thex“specialized
. . . 2

interest of the inquirer or narrator,“ l and "the presence
of specialized efforts, interests, or intentions of the

22 We have already discussed this feature of act-

agent."
descriptions sufficiently in Chapter-I, section 4 ("Descrip-
tion and Purpose").
D'Arcy's last two theses are particularly important
for the problem at hand,.namely, whether and, if so, to
what extent actions may be correctly redescribed~in terms
of their consequences. The second is-stated as follows:
Thesis Two: The term which denoteSathe-act,
~in the-description.of«argivenxincident, may
' often be elidedtinto the termlwhich~denotes
‘a-consequence-Of that act: 'doing X with the

consequence Y', may often be re-desoribed
simply as 'doing Y'.23

 

We have already granted that this claim holds for an entire
host of cases. For example, if A shot B and thereby killed
him, we may say (simply) that 'A killed B'; if I exert
pressure on the door with my hand and thereby close it,

we may say (simply) that 'I closed the door,‘ and so on.

We are already familiar with this claim, so we need not
detain ourselves at this point. However, one feature of
this sort of redescription of acts in terms of their con-

sequences should be emphasized. In cases such as these

99

where we redescribe an act X in terms Of its consequence(s)
Y, we do so by saying simply that"A did Y'; that is, there
is no reference in the redescription, direct or indirect,
to the act X. The important features-of cases of rede-
scriptions of this sort are summarized by D'Arcy:

We may . . . Often elide one possible descrip-
tion, the term X, into another term Y, where
(1) Y is the result or consequence of the
agent A's doing X; (2) A is nevertheless said
to be doing Y . . . (3) the elision is so
complete that Y gives no hint of the specific
nature Of X.

Having said this, D'Arcy then proceeds to formulate the
question with which we have been~dealingwthroughout this
chapter. In his terminology, given that

a is the number of relevant elements comprised
in the term which-is-used to denote the act, a
the number Of relevant elements comprised in
the term used to denote the consequences, and
E the number of relevant elements comprised in
the description of the entire episode, so that

2=:-c

then while E is of course a constant for a
given episode, a and g are variables. But now,
is this true for all values of a and a, or is
their range of possible values restricted?25
Employing less technical, though not necessarily less pre-

cise terms, let me restate D'Arcy's question in a form with

which we are already acquainted: Given that acts may often

100

be correctly redescribed in-terms of their consequences,

is this possible in all cases, or are=therercases in which
such redescription must be excluded?~ Feinberg's "accordion
effect" thesis may be read as a categorical,-affirmative
answer to D'Arcy's question; yes, this variability holds
for all values of a and a, the range of possible values for
a and g is not restricted. In other words, it is, in prin-
ciple, always possible to "squeeze the act down to~a min-
imum" or else "puff it out to-include its~consequences,"
and therefore it is, in principle,-always possible to re-
describe an act in terms of its consequences.

It should be clear that I am forced to answer this
same question in the negative. The range of possible
values for a and 2.15 not unrestricted; that is, we are app
always free to redescribe an act in terms of the conse-
quence(s) of that act. D'Arcy too gives a negative-answer
to this question; in his own words, "there are values above
which a does not extend; that is, there are some act-terms
which can app be elided into the corresponding consequence-
description."26

We will begin our discussion of D'Arcy's reasons

for his View presently. For my own part, independently of

101

any considerations which D'Arcy brings forward, I have thus
far specified at least some "values above which a does not
extend"; simply stated, I have discussed one type of puta-
tive redescription of acts in terms of consequences which

must be ruled out as false and a fortiori incorrect in the

 

strong sense. Specifically, any redescription of an act in
terms Of the consequence(s) of that act which violates the
"intervening agent" principle is galaa. We have, therefore,
completed half the task we set ourselves at the outset of
this chapter; that is, we have succeeded in Showing one of
the ways in which Feinberg's "accordion effect" thesis

fails as a general claim concerning act-descriptions. For
redescriptions of acts in terms of their consequences which
violate the "intervening agent" principle are incorrect in
the strong sense Of being false. We are now ready, fi-
nally, to turn our attention to a discussion of the weak
sense of incorrectness of which Smart's case of the southern
sheriff ('The sheriff saved four lives') and Feinberg's
case of Smith Opening the door and startling Jones ('Smith
killed Jones') are instances. I want to argue that these
sorts of redescriptions, while true, are nevertheless in-
correct in the weak sense insofar as they are in an impor-

tant sense misleading.

102

It is in the context Of accounting for what I have
been calling the weak sense of incorrectness that D'Arcy's
third and final thesis is relevant:

Thesis Three: Certain kinds of acts are of
such significance that the terms which denote
them may not, special contexts apart, be
elided into terms which (a) denote their
consequences, and (b) conceal, or even fail
to reveal, the nature of the act itself.2T

 

Typical examples [of such acts] are the acts
of killing, maiming, slandering, torturing,
deceiving, or seriously Offending another
person; betraying or deserting a friend or
an ally; breaking a contract or a promise
or a confidence; stealing or destroying or
spoiling something which the owner, or the
‘community, looks on as precious; sacrificing
“or endangering one‘s own life, happiness,
- good name, health, or property.28

D'Arcy is at pains to point out that the feature of non-
elidability attaches to such act-terms not because they
denote actions which are necessarily morally wrong, but
rather because of their special importance in human affairs:

To put forward this thesis is not necessarily
to imply . . . that there are certain kinds
of act which are always morally wrong, never
permissible in any circumstances . . . . The
point is not that these acts are so wicked
that sound moral judgment must always condemn
them, but that they are so significant for
human existence and welfare and happiness
that they must always be taken into account
before sound moral judgment can be passed on
them; and if they are to be taken into account,
their presence must first be revealed. The
feature of non—elidability applies, therefore,

103

not only to terms such as 'murder', 'mayhem',

and 'betrayal', into which some sort of moral

judgment is built, but also to terms like

'kill', 'maim', and 'deceive', which may be

purely factual and descriptive.29
I would like to add one Observation to these remarks and
suggest one slight correction. First, I would add that the
revealing of the presence of such act-terms is a prereq-
uisite not only for sound moral judgment, but also essen-
tial for a clear and unambiguous statement of what ac-
tion(s), in fact, has or have been performed; in other
words, it is necessary for the description of an episode
to count as correct. This claim will be defended shortly.
Secondly, it does not seem to me that terms like 'kill,‘
'maim,‘ and 'deceive,’ are "purely factual and descrip-
tive."30 Of course, not all instances of killing, maiming,

and deception are wrong. But it seems to me that such

act-types are at least prima facie wrong to the extent

 

that there is at least an initial presumption against such
acts. But this does not affect D'Arcy‘s thesis seriously,
and we can still agree with his central claim concerning
such act-terms:

some elements, or combinations of elements are

such that, whenever they are present, they

verify an autonomous act-definition. They
constitute an instance of some kind Of act

104

whose description, special contexts apart,
cannot be absorbed into some other: that
'other' remains a consequence of the act,
not a part of it: extrinsic, not intrinsic,
to it. They constitute a 'case' of morally
significant action.31

For brevity and convenience, let us henceforth refer to
D'Arcy's Thesis Three as the "elidability criterion."

The upshot of Thesis Three or the "elidability
criterion" is to restrict the application of the second
thesis; in other words, Thesis Three amounts to the claim
that, although we may often correctly redescribe an act
in terms of its consequences, this is not always possible:

This thesis, then, is supplementary to the

theses that several alternative descriptions
Of a given act are possible, and that one may
Often, in analysing a given incident draw the
line between 'act' and 'consequences' now at
one point, now at another. It holds that in

certain cases there is One point at least at

which an 'act'/'consequences' line must be
drawn.32

It is our task now to discover precisely which sorts of
cases of redescription are excluded by the elidability
criterion and, more importantly, the reasons for their
exclusion.

We can begin immediately by disposing of Smart's
illustration. Since 'killing' is a "typical example" of

that type of act which cannot be elided into a term

105

denoting the consequence(s) of that act, we must on D‘Arcy‘s
criterion deny that we may correctly redescribe the sher-
iff's act simply as 'saving four lives‘ without any refer-
ence to the means he employed to save those lives (viz.
by killing another man). We may, of course, employ a fuller
description of the episode in order to avoid the infelicity
engendered by this description, one which makes explicit
reference to both relevant elements in the episode; we may
say, for example, that ‘The sheriff executed one innocent
manand thereby saved the lives of four others,‘ or 'The
sheriff-managed to save four lives by executing one inno-
cent Negro,‘ or so on.

SO far so good; the elidability criterion does
enable us to handle the case of the southern sheriff.
Given that certain types Of act-terms of which ‘killing‘
is an instance are not elidable into terms which denote
the consequences of those acts, we are able to rule out
the redescription ‘The sheriff saved four lives' as in-
correct. But we may still legitimately inquire yhy just
those act-terms mentioned by D'Arcy are non-elidable.
D'Arcy‘s answer is stated in terms of the human signifi-

cance which attaches to such acts. With respect to killing,

106

for example, he tells us that "taking human life, we feel,
is an act of such significance that one cannot elide its
description into a term which denotes its consequence, or
an end to which it was a means, unless that term makes
clear that this was the means used."33 This is an instance
of one of those cases in which the act/consequence line
THEE be drawn.

D'Arcy has been criticized in this respect for
making the question of elidability or non-elidability of
various act—terms a contingent matter of fact. Just which
act—terms are or are not elidable into consequence-terms,
some claim, is wholly relative to the time-span and loca-
tion in which we are interested. Perhaps ‘killing' and
'maiming‘ are of such moral significance at tl or at
place pl that they may not be elided into terms denoting
their consequences; but there is no guarantee that these
two act-terms will enjoy the same privileged status at
time t2 or in locale p2. Perhaps in some future age or
in some other society 'killing' and 'maiming' will be of
no more significance than ‘closing doors‘ or 'playing
badminton‘; however unlikely or inconceivable such a state

Of affairs may appear, there is no logical impossibility

107

in claiming that such a state of affairs could Obtain.
Therefore, it is claimed, since the notion of elidability
is so fraught with this sort Of relativism, D'Arcy's thesis
is not well-founded. Joseph Margolis, for example, advances
precisely such considerations as grounds for rejecting the
elidability criterion. He attacks D'Arcy on the grounds
that his position is tantamount to "moral essentialism,"
a view which holds that

given man‘s nature, it is necessary to admit

as morally relevant the characterization of

all acts under D‘Arcy's non-elidable act-

descriptions and it is even conceptually

wrong to ignore or obscure or eliminate any

of these categories in the context Of moral

judgment.34
To counter this conception, Margolis invites us to imagine
a society in which the only means of succession to the
throne is via a successful murder attempt; in such a
society, he claims,

the consequence of succeeding to the throne

could, reasonably, be elided with the de-

scription Of killing--our man could, under

the circumstances, be said to have per-

formed the act of establishing his legiti-

mate title tO the throne.35
And presumably the same considerations apply mutatis

mutandis to each Of the remaining act-terms which D'Arcy

considers non-elidable. Given conceivability as the

108

criterion Of possibility, and given that we can always
imagine or conceive of a society in which . . . , it
follows that none of the act-terms which D‘Arcy considers

non-elidable are necessarily of such dire Significance for

 

human existence, welfare, and happiness. According to
Margolis, what this argument entails is

the possibility of radical reforms and
changes in the most persistent and seem-
ingly fundamental practices of human
society. If D‘Arcy‘s thesis is false,

if terms denoting morally relevant acts
and their consequences may, in principle,
always be elided, then no moral categories
of the relevant sort are unalterable.
Those terms like ‘murder,‘ 'mayhem,‘ and
‘deceit,’ that have "some sort of moral
judgment" built into them, may always be
detached from such judgments and redefined
as "purely factual and descriptive." And
those terms like 'kill,‘ 'maim,‘ and‘de-
ceive,‘ . . . may be freed from any aroma
of moral import whatsoever; they may, in
principle, cease to be relevant . . . "for
reaching a sound moral verdict."36

 

 

 

80 runs the argument against D'Arcy‘s elidability cri-
terion, an argument which depends for its force on the
notion of what is and what is not logically possible.
Although I too find D'Arcy's account of elidability
in terms Of degree of human or moral significance unsatis-
factory as it stands, I find Margolis‘ reasons for reject-

ing it even more unsatisfactory. Of course, it is not

109

logically impossible that certain acts which we now find of

 

great significance may not have such importance at some
later time. But this is to admit--what? That-—given that
human beings are what they are, with the languages that
they do in fact have, with the terms that they do in fact
employ, which terms do in fact have the meanings they have--
act-terms like 'killing' and 'maiming' do have particular
significance in the normal conduct Of our lives. But how
can this in any way be construed as a telling objection to
D‘Arcy‘s claims? Who ever claimed that it is necessary
that words do have the meanings which they do in fact have,
or that concepts cannot or do not sometimes change? It is
logically possible that we could in the future collectively
stipulate that the term 'bachelor‘ is to be used in such a
way that it has no reference to a person's marital status
or membership in the male of a Species. Of course, this
would require a change in our present concept of ‘bachelorfl
But my point is that this would necessitate no more change
than would be effected by a change in our present concept
of ‘maiming' if, at some future date, 'maiming‘ is no
longer considered an instance Of a type Of act which is

regarded as prima facie wrong. Of course, if such momentous

 

110

changes did in fact come about, we would be forced to amend
our present catalogue Of non-elidable act-terms. But so
what? We can with as much confidence right now assert that

'If Smith maimed Jones, then Smith did something ppima facie

 

wrong,‘ as we can now assert that ‘If Smith is a bachelor,
then he is unmarried.‘ And what interests us here are the
meanings Of and interrelationships between the various act-
terms which we do have at our disposal and which we do
employ in characterizing the things we do. In short I do
not feel that Margolis' argument poses any serious threat

to D‘Arcy‘s elidability criterion.

(v) A Modified Account
of Elidability

 

 

The foregoing Observations alone, however, are not
sufficient for vindicating D'Arcy‘s position. As remarked
above, I too find his account of elidability in terms of
degree of moral significance unsatisfactory. I do not
deny the truth of what D‘Arcy is maintaining. Assuredly,
acts like maiming and killing are of especial importance
in human affairs; this we must concede. Rather, I feel

that D‘Arcy has not gone far enough in exploiting the

111

implications of his claims concerning elidability for their
explanatory power. Given our concept of "implication," we
can, I believe, make a real start at supporting D'Arcy‘s
claims and strengthening his position.

Let us begin with D'Arcy. Suppose that act-term X
is one Of those which is not elidable into Y, a term denot-
ing the consequence(s) of X. We may now ask, ‘Why is X not
elidable into Y?‘ And D‘Arcy will answer, 'Because X is
one of those act-terms which is of particular human signif?
icance or moral import.‘ But, we may now ask, why are act-
terms like X which admittedly are of particular significance
in our lives such that they may not be elided into corre-
sponding consequence-terms? To this question D'Arcy himself
gives no answer. But, armed with our notion Of "implica-
tion" we can, I believe, provide a reasonable account of
the prohibition: Act-terms like X which are of particular
human significance are not elidable into consequence-terms
because such redescription glves rise to infelicitous

"implications"; that is, such redescription legitimately

 

gives my audience to understand that the episode in ques-
tion is characterized by various features which, in fact,

do not or may not Obtain.

112

In order to clarify and support this position, let
me begin by applying this analysis to our previous illus-
trations. Recall that we were able to rule out the de-
scription, 'The sheriff saved four lives‘ as incorrect
(even though true) on D'Arcy's elidability criterion, since
this description elides the non-elidable act-term 'killing‘
into a phrase which denotes its consequence. We are now in
a position to go one step further and explain yhy such
elision is impermissible. To describe the sheriff's act
simply as ‘saving four lives‘ is misleading in the sense
that my audience has not been given the "whole story." My

failure to mention the prima facie wrong means employed by

 

the sheriff (viz. killing an innocent Negro) has various
consequences of its own. My audience may, for example,
legitimately assume that the episode contained no elements
which they might otherwise regret under different circum-
stances. After all, would it not have been even better if
the sheriff could have managed to save those lives without
employing the means he found himself forced to employ? To
say simply that ‘He saved four lives‘ disguises the fact
that the sheriff was forced to choose the lesser of two

evils, that the episode is not wholly cause for rejoicing

113

because of the presence Of this one tragic element. Of
course this is not to say that our audience would or ought
to alter their moral assessment of the sheriff‘s act or to
withhold their praise Of him for performing it if they were
given the "whole story." It may very well be that it took
a person of strong moral fiber and exceptional courage to
take the decisive action which the sheriff did take in
those difficult circumstances. In other words, our evalua-
tion of the sheriff and his act would not necessarily change
if we were given the "whole story" via an expanded act-
description which included reference to the killing (e.g.
‘The sheriff at least managed to save four lives; unfortu-
nately, he could accomplish this only by executing an inno-
cent Negro‘). In other instances, of course, we pay find
that we wish to alter our evaluation once we are given the
"whole story" including all of the relevant circumstances.
But even here, once the (or a) correct description has been

fixed on, it is the evaluation—-not the description itself--

 

which is altered. In this context, D'Arcy approvingly
paraphrases Bentham: "Where there are 'modifying‘ circum-
stances . . . it is the evaluation which may be modified,

not the description."37

114

I think my position gains even more plausibility
when we apply the analysis I am urging to Feinberg‘s case
in which Jones Opens the door and startles Smith and, as a
result, Smith suffers a heart attack and dies. Remember
that, according to the "accordion effect" thesis, we may
redescribe the act in question simply as 'Jones killed
Smith.‘ We can now rule this description out as incorrect
since it "implies" things which are not true. Indeed, with
respect to thise case, Atwell remarks that "curiously, one
can hardly imagine a better counter—example to the accordion
effect thesis."38 As for the infelicitous implications of
this action-sentence, let us attend to what he says about
this example:

For, first of all, to say that Jones killed
Smith is to suggest to one's audience a number
of things which no one, I think, wishes to
suggest: (1) that Jones did something prima
facie wrong, assuming that it is prima facie
wrong to kill people; (2) that Jones thus did
something he should not have done; (3) that

he could have avoided killing Smith, assuming
that 'should not have' implies ‘could have
avoided‘; and finally, (4) that Jones must be
prepared to defend, by justification or excuse,
his "action". Since we wish to suggest none
of these, we should not ascribe the act of
killing Smith to Jones--in spite of our will—
ingness to ascribe Smith's death to Jones‘s
opening the door. In other words, merely
because we can trace someone's actiOn to
certain causal consequences, it does not

 

 

 

115

automatically follow that we may incorporate

those consequences into a "broader" descrip-

tion of the person's action. Furthermore, a

man can regret or take pride in his actions.

But if Jones were to assert in deep agony "I

killed Smith", we would surely regard his

behavior, i.e., his asserting this, as irra-

tional and would try to dissuade him Of such

a claim; and if he were to assert it in a

boastful manner--for some reason--we would

dismiss his claim as wholly unwarranted.39
The type of infelicity engendered by the action-sentence
‘Jones killed Smith‘ in Feinberg's illustration is spelled
out nicely here by Atwell. In uttering this sentence, I
give my hearer(s) to understand (1)-(4), none of which by
hypothesis obtain in these circumstances. Once again the
infelicity could be avoided by "filling in" the relevant
details, that is, by expending or qualifying the descrip-
tion in one or more appropriate ways. We may wish to re-
late the entire episode in toto. But we could just as well
say, for example, that 'Jones killed Smith unintentionally/
accidentally.‘ By thus incorporating the qualifying phrase
‘unintentionally‘/'accidentally,' we thereby "cancel out"
the infelicitous implications of our original action-
sentence. According to Atwell, "in such a case we will

canalify the action-sentence 'Jones killed Smith‘ with a

'term like 'accidentally' or 'unintentionally', thereby

116

evading the implicit suggestion that Jones is somehow at

fault."40

 

Although Atwell casts his analysis in terms of
"implicit suggestion," it is clear that this is substan-
tially the same as my notion of "implication." Before pro-
ceeding to examine the various ways in which action-sentences
may fail to be correct in virtue of what they "imply," I
would first like to show why my and Atwell‘s analysis, al-
though it often results in the same substantive conclusions
which are forthcoming on D‘Arcy's elidability criterion,
nevertheless has more explanatory power and is thus capable
of dealing with cases which the elidability criterion alone
cannot handle. The illustration just considered provides
a case in point. We wanted to rule out the action-sentence
'Jones killed Smith‘ as incorrect because it "implies"
or suggests things which we do not with to "imply" or
suggest under the circumstances. But the elidability
criterion itself cannot rule out this act-description for
the simple reason that the act-term in question, namely,
'closing a door,‘ is not one presumably which is of such
special significance that it cannot be elided into a

consequence-term.41 In short, then, I am suggesting that

117

we regard the elidability criterion as one species of our
analysis. Whenever the elidability strictures are violated,
so too will our requirement be violated; that is, whenever
one of D'Arcy's non-elidable act-terms is elided into a
corresponding consequence-term, the resulting description
(will be incorrect at least in the sense Of being misleading.
However, since the scope of our analysis is broader than
D‘Arcy‘s, the converse will not hold; that is, it will not
always be the case that whenever the requirements of our
analysis have been violated that the elidability criterion
will necessarily thereby be violated.

The elidability criterion gives rise to another
sort of difficulty, one to which D‘Arcy himself is sensi-
tive. Not only are there cases in which one may not elide
an act-term into a consequence-term, but there are also
cases in which such elision is necessary for avoiding
"implications" which may be misleading. Recall once again
the incident involving the southern sheriff and D‘Arcy‘s
list of possible descriptions. Given the events involved
in this scenario, to gall to elide ‘He tensed his fore-
finger‘ into one or more of its consequences (‘pulled the

trigger,‘ ‘Shot a gun,‘ 'shot a man,‘ ‘killed a man‘) would

118

be just as misleading as saying (simply) that ‘He saved
four lives.‘ In this context D‘Arcy remarks that "in terms
Of the twelve contemplated descriptions of the Sheriff‘s
act, Number 1 ('He tensed his forefinger‘) usually Should
be elided into one of the latter descriptions, e.g. Number 9
(‘He shot a man‘) . . . . Tensing one‘s forefinger is
simply part of the act of shooting, or shooting a man

. . . ."42 I do not wish to dispute what D'Arcy says here;
all I am claiming is that this does not follow from his own
statement of the elidability criterion, that independent
reasons must be advanced to support this claim, and that
our analysis does in fact provide the grounds needed. For
to say in this case simply that 'He tensed his forefinger'
would be to "imply" or suggest that what he did was of no
more significance than snapping one‘s fingers or blowing

one‘s nose. And this, I maintain, is something we would

not want to suggest in those circumstances.

(vi) Conditions for Re-
describing Acts in Terms
ijConseguences

 

 

It is now time to turn to a consideration of cases

With a View to discovering the grounds upon which we may

119

determine what sorts of redescriptions of acts in terms of
their consequences are permissible and what sorts are to
be ruled out as incorrect. The first condition I will sug-
gest is really a restatement of D‘Arcy's elidability cri-
terion in slightly different terms; it will enable us to
account for the type of infelicity of which Smart‘s de-
scription of the sheriff‘s act as ‘saving four lives' is
an instance. Let me begin by stating the condition, and
then we will see how it applies to the case at hand.
Cl We may not redescribe an act, X, simply
in terms of its consequence(s), Y, when
X denotes an act-type which is pplpa

facie wrong and Y denotes an act-type
which is ppima facie right.

 

In Smart's illustration, then, we may not describe the
sheriff‘s act simply as ‘saving four lives‘ since this
elision violates Cl; that is, it violates our prohibition
against eliding an act-term which denotes a prima facie
wrong act (viz. ‘killing‘) into a consequence-term denoting

an act-type which is ppima facie right or commendable (viz.

 

'saving lives‘). And the reason for this is that to elide
an act-term in this way is to suggest or give our audience
to understand a number of things that none of us are pre-

pared to accept, things which we do not wish—-or should not

120

wish--tO suggest to our audience. We have already discussed
the infelicities engendered by the act-description 'The
sheriff saved four lives' in Smart‘s illustration.

The force of Cl will figure importantly in many dis-
cussions of ethical issues. For example, Cl dictates that
we may not describe an act of euthanasia simply as 'reliev-
ing the patient's distress,‘ nor may we describe an act Of
therapeutic abortion simply as 'saving a mother‘s life,‘
even though it may be papa that a patient‘s distress has
been relieved and that a mother‘s life has been saved,
respectively. Both descriptions are misleading in virtue

of eliding a prima facie wrong act-term (in both cases,

 

the term 'killing') into consequence-terms which denote

acts which we regard as paima facie right or commendable

 

('relieving a person‘s distress‘ and 'saving a person‘s
life,‘ respectively). In both instances we have not been
given the "whole story"; the descriptions under consider-

ation fail to mention how these prima facie right Objec-

 

tives have been accomplished, and it is this failure which
engenders the infelicity.

In general, to redescribe prima facie wrong X

 

solely in terms of prima facie right Y suggests to one‘s

 

121

audience that something good, praiseworthy, or commendable
has been accomplished without qualification, that nothing
was amiss in the episode, nothing which otherwise may be
regarded as unfortunate or regrettable-~even if justifiable.
And these, I maintain, are precisely the sorts of things
one ought to avoid suggesting to one‘s audience in such
cases (unless, of course, his intention is to deceive that
audience).

But what about the converse of Cl; may we or may we

not redescribe the prima facie right act-term, X, in terms

 

of prima facie wrong consequence-term, Y? Although D‘Arcy

 

restricts his list of non-elidable act-terms tO those which

denote acts that we may say are paima facie wrong, at least

 

to the extent that there is a presumption against such acts
as killing, deceiving, and so on, is it not plausible that
the same or Similar considerations apply as well to act-

terms which denote prima facie right acts, or those for

 

which there is a presumption in favor of their performance?
I think that this is plausible, and I would like to suggest
that we extend our modified elidability criterion to such

cases as well; after all, these prima facie right acts are

 

also important for human existence, welfare, and happiness.

122

I want to argue that in general one may not elide

a prima facie right act-term into a consequence—term which

 

characterizes a paima facie wrong act. Suppose, for ex-

 

ample, that my dear friend Jones asks me about his wife‘s
extramarital affairs. He insists that I tell him the truth,
no matter how much it might hurt, and I feel obliged to do
SO. As a result, Jones's marriage goes on the rocks. Al-
though my telling Jones the truth about his wife may to
some extent be "responsible" for the breakup, I do not be-
lieve that we would wish to elide my 'telling Jones the
truth‘ into a description of its consequence by saying that
'I broke up the Jones's marriage.‘ This description sug-
gests that I am the culprit, that my role in the affair was
blameworthy; it "implies" that I am as much at fault as
Smith who, in fact, has been consorting with Jones‘s wife
when Jones is away. But, by hypothesis, this is not the
case, and we certainly would not wish to suggest that it is.
On the basis of this illustration alone, one might

be tempted to conclude immediately that a prima facie right

 

act-term X may never be elided into a prima facie wrong

 

consequence-term Y. But this, I feel, would be too hasty.

Recall that when I introduced this discussion I claimed

123

only that such elision is in general impermissible. This

 

is because there does appear to be one factor which may

affect our willingness to redescribe prima facie right act,

 

X, in terms Of prima facie wrong, Y, namely, the factor of

 

the agent's intention in doing X in the first place. Con-
sider again the adulterous Ms. Jones. Suppose that her
lover, Smith, goes out of his way to inform poor unsuspect-
ing Mr. Jones of his wife's illicit affairs in an explicit
attempt to break up their marriage. Smith does not like

the thought Of sharing Ms. Jones with her own husband, but
would rather have her to himself; and this seems as effec-
tive a means as any under the circumstances to accomplish
his end. Given that Smith had every intention of breaking
up the marriage (and given that we know or have good reasons
to believe that this was, in fact, his intention), I do not
think that we would--or should--hesitate to describe Smith‘s
informing Mr. Jones of his wife's extracurricular activities
as 'Smith broke up Jones‘s marriage‘ (or ‘Smith tried to
break up Jones‘s marriage,‘ in the event that his intention
is not realized and his attempt fails). In doing the prima
facie right X (telling the truth), Smith intended to accom-

plish the paima facie wrong Y (viz. breaking up a marriage).

 

124

In fact, given that this was Smith‘s intention, we may very
well find ourselves reluctant to describe his act in the
first instance simply as 'telling the truth‘; rather, we
might feel compelled to expand this in some way in order to
convey the "whole story," and what better way of doing this
than by eliding it into the term denoting the consequence?
Given this proviso regarding intention, we may formulate
the second condition governing the redescription of acts in
terms of their consequences as follows:
C2 In general, we may not redescribe an act,

X, simply in terms Of its consequence(s),

Y, when X denotes an act-type which is

prima facie right and Y denotes an act-

type which is paima facie wrong, unless
the agent intended to do Y in doing X.

 

 

Here when we say that the agent intended to do Y in doing
X, this entails that Y is something aimed at or wanted by
the agent (this will be distinguished from other species
of intentional action in Chapter III).

Fortunately, however, not all Of the things we do
during the course of our lives are of such moral signifi-
cance; we have already had occasion to emphasize this
feature Of actions in our discussion of the Hart-Melden
thesis of ascriptivism. Many, in fact most, of the things

we do and many of the terms we use to denote acts or

125

act-types have no moral or quasi—moral implications at all.
SO what are the relations between morally "neutral" act-
types and their consequences, and what bearing do these re-
lations have on the problem Of redescribing actions? Taking
as our basic statement-schema 'A did X and thereby caused
Y,‘ let us investigate some of the possibilities to deter—
mine under what conditions we may and under what conditions
we may not say simply that ‘A did Y.‘

Let us begin by considering a case of a morally

neutral act X which has the prima facie wrong/bad conse-

 

quence Y. Consider once again our favorite example from
Feinberg. Jones closes the door; unbeknownst to him, Smith
is standing on the other side. Smith is startled by the
noise of the door closing, suffers a heart attack, and dies.
Here we have a case Of the desired sort, a case of a morally
neutral act (viz. closing a door) resulting in an unfortu-
nate incident (viz. Smith's death). Nevertheless, we have
seen, we will not countenance describing this act simply as
'Jones killed Smith' since this would suggest to one‘s
audience a number of things we would not wish to suggest
given the circumstances. This case is very similar to those

ruled out by C1. Just as we cannot elide a prima facie

 

126

wrong act-term into a consequence-term denoting an act

which we regard as prima facie right, so too we will not

 

in general allow a morally neutral act-term to be elided

 

into a consequence-term which denotes an act that we regard

as prima facie wrong, as in the present case. However, the

 

qualification 'in general‘ is very important here. For
there do seem to be (at least) two considerations which
might affect our willingness or unwillingness to make this
sort of elision. The first of these parallels the quali-
fication concerning intentions that was appended to C2.

That is, if the prima facie wrong act denoted by the

 

consequence-term Y is in fact intended by the agent in
doing the morally neutral X, then we may very well be in-
clined to say simply that ‘A did Y.‘ In the present case,
for example, suppose that Jones aaaa know that the weak-
hearted Smith is on the other side of the door and, fur-
thermore, that Jones does intend to kill Smith by startling
him. It seems to me that if we know or suspect Jones‘s
intention in these circumstances, we would be quite willing
to say that ‘Jones killed Smith.‘ So the presence or ab-
sence of the relevant intention may affect our willingness
or hesitation to redescribe acts in terms Of their conse-

quences in such cases.

127

Secondly, sometimes an undesirable consequence may
be an expected result or "natural consequence" of some act
which, under a given description would, under most circum-
stances, be regarded as morally neutral Or indifferent.
D'Arcy suggests a case of this kind: "if a passenger
speaks to a bus driver and thus distracts him, we may say
Simply that he distracted him."43 Now it may seem that we
would be dealing rather high-handedly with English usage

were we to claim that distracting bus drivers is an instance

of paima facie wrong act-types. Nevertheless, there is a

 

presumption against distracting bus drivers; after all,
there are signs which read 'DO NOT DISTRACT THE DRIVER,‘
and we do admonish and blame people for doing so. In any
event, it does seem that in a case like this we are willing
to redescribe the ‘talking' in terms of its consequences,
‘distracting.’ The idea behind our willingness to rede-
scribe this act in this way is vindicated by our ordinary
talk about actions. "It is a presumption of law, logic,
and common sense," said Sam Ervin while berating one of the
Watergate principals, "that a man is presumed to have in-
tended the natural consequences Of his act." D‘Arcy for
his part speaks in terms Of the "anticipatibility" of such

consequences:

128

we elide the act-term X into the consequence-
term Y when Y is the kind of consequence that
may be anticipated from an act of the kind X:
when, that is, Y is naturally, on general
grounds, anticipatible by the agent or in-
formed Observer.

 

On the other hand, one does not make the

elision if the consequence is completely

unexpected, even by the observer: ‘doing

X with the consequence Y‘ is not re-described

as ‘doing Y‘, if there is no reason to think

that this sort of act, X, will have this

sort of consequence.44
Notice, however, that the consequence need not be antici-
pated by the agent himself. In our illustration, perhaps
the passenger did not know that his talking would distract
the bus driver, and that he had no intention of doing so.
Nevertheless, since the distraction is, on general grounds,
an anticipatible consequence of talking to bus drivers,
we may very well describe the passenger‘s talking simply
as ‘distracting the bus driver.‘ This notion of anticipat-
ibility also helps to account for the infelicity Of the
act-description ‘Jones killed Smith' in Feinberg‘s case.
Assuming that Jones did not intend to kill Smith, we may
not redescribe his 'closing the door' simply as ‘killing
Smith‘ since killing persons is not, on general grounds,

an anticipatible consequence Of closing doors. With these

qualifications concerning (1) intentions, and

129

(2) anticipatibility of consequences in mind, we may now
formulate a third condition governing the redescription of
acts in terms of their consequences:

C3 In general, a morally neutral act-term X
may not be elided into a consequence-term
Y which denotes a paima facie wrong act
unless, either (a) the agent intended to
do Y in doing X, and/or (b) Y is the kind
of consequence that may be anticipated,
on general grounds, from an act of the
kind X.

 

TO violate this condition, that is, to redescribe A‘s prima

facie neutral act X in terms of its prima facie wrong

 

consequence-term Y by saying simply that 'A did Y‘ would
be to suggest that A had done something wrong, that he is
or may be blamed for doing Y; and in the absence of the
two qualifying circumstances, these would be things we
would not wish to suggest. For instance, suppose Jones‘s
closing the door caused the vase to fall Off the table and
break. Breaking vases is certainly not a generally anti-
cipatible consequence of closing doors. Assuming that
Jones did not intend to break the vase, we would not wish
to say in such a case simply that 'Jones broke the vase,‘
even though we would not hesitate to attribute the break-

ing of the vase to Jones‘s closing of the door.45

130

Consider now a case similar to those covered by C3;
that is, a case in which we have a morally neutral act X

with prima facie right, good, commendable, or felicitous

 

consequence Y. May we describe such a sequence Of events
simply as ‘doing Y'? Suppose Jones returns to his home one
night; unbeknownst to him, there is a would-be burglar
prowling about in his garage. Jones flips on the light
switch; as a result the burglar is alerted to the fact that
the owner has returned and immediately flees from the scene.
Jones has succeeded in preventing a burglary from occurring,
or at least the burglary was prevented as a result of what
Jones did. Now ought we allow the elision of Jones‘s 'turn-
ing on the light‘ into its consequence, ‘preventing a
burglary‘; may we say simply in this case that 'Jones pre-
vented a burglary‘? Although we may be quite willing to
ascribe the prevention of the burglary to Jones's returning
home and turning on the light, and might readily be pre-
pared tO say that that 'Jones accidentally (or unintention-
ally) prevented the burglary,‘ I nevertheless do not think
that we should describe what Jones did simply as ‘preventing
a burglary.‘ It might be objected immediately that ‘pre-

venting burglaries‘ is something that one can accomplish

131

unintentionally as well as intentionally; and with this I
agree (see Chapter III, section 2). All I am claiming is
that if—-as in the present case—~it (preventing the bur-
glary) is done unintentionally, then we must indicate this
in our description in order to avoid the risk of misleading
our audience. To say simply that ‘Jones prevented a bur-
glary‘ would be to suggest that Jones did something for
which he deserves our praise, even suggesting perhaps that
he jeopardized his own safety to accomplish the task; since
we do not normally praise people for "doing good acciden-
tally," these are things we would not wish to "imply" or
suggest. On the other hand, of course, if Jones knew or
suspected that there was a would-be burglar lurking in the
garage, and he intended to scare him Off by turning on the
light, then we would not hesitate to say that 'Jones pre-
vented a burglary.‘ But given that ‘preventing burglaries'
is not, on general grounds, an anticipatible consequence of
‘flipping a switch' or ‘turning on a light,‘ we would not
be prepared to say of Jones that he prevented a burglary
unless this were his intention in turning on the light.

If this is correct, then we may formulate our next condi-

tion as follows:

132

C4 In general, a morally neutral act-term,X,
may not be elided into a consequence-term,
Y, which denotes a prima facie right or
praiseworthy act unless the agent in-
tended tO do Y in doing X.

 

Notice that in C4--unlike C3--we do not permit the elision
even if Y was an anticipatible consequence Of X. This is

so because if the agent did not know that Y was an antici-
patible consequence of X and furthermore did not intend to
do Y, we do not wish to praise him for doing good acciden-

tally. In a case covered by C3 in which prima facie wrong

 

Y is an anticipatible consequence of X, we would say that
the agent should have known better; the passenger should
have known that talking to the bus driver could serve only
to distract him. But if in a case of the present sort

where A did not know that paima facie right Y is an anti-

 

cipatible consequence of X, and did not intend to do Y in
doing X, we would not go on to praise him for doing X,
adding that he should have known that it would have the
felicitous consequence Y.

Consider finally a case in which the statement
schema ‘A did X and thereby caused Y‘ is filled out by

replacing both X and Y with terms denoting prima facie

 

neutral act-types. Take one of our original examples:

133

suppose I exert pressure on the door with my hand (X) and
thereby close the door (Y). We have already seen that
there is no hesitation in this case to redescribe my doing
X solely in terms of its consequence Y by saying simply
that ‘I closed the door' (i.e. ‘A did Y‘). In this case,
then, it is permissible to redescribe one morally neutral
act X in terms of its morally neutral consequence Y. But
one must resist the temptation to conclude immediately that
this sort of redescription is permissible in all such cases.-
Suppose that the impact of my closing the door caused the
activation of the sprinkling system in the room. Although
we perhaps would not hesitate to say that my closing the
door caused the sprinkling system to activate, would it not
be misleading to say simply that 'I activated the sprinkling
system'? I suggest that it would be misleading, and that
this is so because, among other things, ‘activating
sprinkling systems‘ is not a generally anticipatible
consequence of ‘closing doors‘ in the way that 'closing
doors' la a "natural consequence" of exerting pressure on
them with one‘s hand. If, on the other hand, I knew that

my closing the door would activate the sprinkling system,

and if I intended to do precisely that, then (given that

134

those who are Observing the spisode know or suspect my
intention), we might very well be willing to describe my
act simply as ‘activating the sprinkling system.‘ Once
again we see the important bearing of the agent's inten-
tions and the notion of the anticipatibility Of consequences
in determining how willing we are to describe and redescribe
various sorts of actions in certain ways. These factors
figure importantly in the statement Of our final condition
governing the redescription of acts in terms of their con-
sequences:
C5 In general, a morally neutral act-term, X,

is elidable into a morally neutral

consequence—term, Y, to the extent that

Y is an anticlpatible consequence of X;

if Y is not an anticipatible consequence

of X, then X may not be elided into Y

unless the agent intended to do Y in doing
x.

 

 

What I have tried to do in this section is to indi—
cate some of the ways in which a redescription of an act
in terms of its consequence(s) may fail to be correct even
though the description itself may be true. I have tried
to do this in terms of indicating what certain types of
descriptions "imply" or suggest to one‘s audience. If a
given act-description legitimately gives one's audience

to understand things which one is not prepared to accept

135

or assert, then that description is to be ruled out as in-
correct in the weak sense of being misleading. We consi-
dered a number of cases with a View to determining whether
there are any general sorts Of considerations to guide us
in our efforts to determine whether a given redescription
of an act in terms of its consequences is correct and thus
"permissible," or is to be ruled out as misleading and
thus incorrect in the weak sense. Conditions C1-C5 repre—
sent the fruits of our labor; each of these in effect
states a condition the violation of which serves to defeat

a true act-description‘s claim to correctness.

Summary

Thus far in this chapter we have addressed our-
selves tO the problem of redescribing acts in terms of
their consequences. Having granted that one may frequently
redescribe an act solely in terms Of the consequence(s) of
that act, we investigated whether, as some claim, such
redescription is always possible in principle. In section
(i) I showed that some such redescriptions fail to meet

even the minimally necessary condition for correctness,

136

namely, truth; and in section (ii) I suggested an explana-
tion to account for this failure in terms of the "inter-
vening agent" principle. In section (iii) we discussed the
problem of redescriptions which, while true, nevertheless
failed to be correct insofar as they are misleading. To
account for this feature, I suggested that we incorporate
the insights of D‘Arcy‘s elidability criterion into a more
general account of a description‘s being infelicitous or
misleading in terms of what a given description "implies"
or suggests to one‘s audience (sections (iv)-(v)). Finally
in section (vi) we turned to an examination of various
cases with a view to determining what sorts of general
conditions, if any, there are to govern the redescription
of acts in terms of their consequences. These (Cl-C5) in
effect specify some of the conditions the violation of
which serve to defeat a true act-description‘s claim to
correctness. Of course, I do not pretend that our dis-
cussion in any way represents an exhaustive analysis or
treatment of the issue; one cannot anticipate a priori the
myriad sets of circumstances or special contexts that may
arise. But I have at least tried to indicate some of what

I take to be the more important considerations which must

137

be taken into account when we are attempting to determine
whether we may or may not correctly redescribe a certain
act or act-type in terms of its consequences. We will have
occasion to expand our analysis and amplify some of our
former claims in the next chapter, in which we will con-
sider the Special role of intentions in the problem of

act-descriptions.

2. Implications for Ethical Theoay

Before closing this chapter, I would like to
briefly examine what bearing, if any, the results we have
reached thus far have for certain fundamental concepts in
ethical theory. In particular, I would like to examine
the effect that our discussion of acts and consequences
has on the traditional distinction between deontological
and teleological ethical theories.

It has been suggested by some that the thrust Of
the 'one action--many descriptions' thesis in effect renders
the distinction between deontological and teleological
theories vacuous or superfluous. Insofar as this distinc-

tion turns on the controversy over the importance of

138

consequences in determining the moral worth of an act, it

is claimed, the fact that an act may be redescribed in terms
of its consequences reduces the controversy to an argument
at cross-purposes. Oldenquist has expressed this View with
force and clarity.46 He characterizes the deontologist as
one who holds that the rightness (or wrongness) of an action
is determined solely by the type of action it is. Although
this may be true of certain extreme deontologists, perhaps
it would be more accurate to say simply that the deontolo-
gist is one who denies what the teleologist asserts; namely,
that the rightness (or wrongness) Of an act is determined
solely by the goodness or badness of its consequences. As
is well-known, adherents of each of these positions must
attempt to deal with apparent counter-examples. On the one
hand, the anticipated consequences of some actions seem to
force one to make occasional exceptions to moral rules. For
example, in Kant‘s famous example of refusing to tell a lie
even though this would certainly result in saving a friend's
life, few Of us are willing to go so far as Kant in never
allowing exceptions to the moral rule ‘DO not lie.‘ On the
other hand, certain types Of actions do seem morally wrong,

no matter how desirable their consequences. Perhaps

139

Stalin's liquidation of the kulaks did promote the general
welfare to an enormous extent; nevertheless, none of us, I
hope, would condone such a pogrom, much less commend Stalin
for executing it with such efficiency, the Principle of
Utility notwithstanding. Oldenquist observes that each
theory attempts to avoid these counter-examples and redeem
itself "by exploiting the reciprocal translatability of
action descriptions and result descriptions,"47 or, to use
the terminology we have been employing, by exploiting the
fact that an act may be redescribed in terms of its conse-
quences. According to Oldenquist, this leads to the fol-
lowing sort Of result:

If an action produced by Jones, such as "a
life being saved," is deemed morally relevant
or "right-making" by the formalist [i.e. the
deontologist], he classifies it as a kind Of
action and hence as the proper subject matter
of a rule. If the same occurrence is con-
sidered morally relevant by the teleologist,
he classifies it as a result and hence as
something that can be good or bad. If our
analysis of the distinction between actions
and consequences is correct, these two meta-
ethical theories are compatible, and because
their importance for ethics has rested on
their sppposed incompatability, each is
trivial.48

 

 

 

 

 

Oldenquist summarizes his claims concerning actions, conse-

quences, and ethical theory very neatly by means of the

140

following argument: we are to "suppose that Jones‘s action
involves (a) the characteristic 'being a lie‘ and (b) the

characteristic 'saving a life,‘ and that both the formalist
[i.e. the deontologist] and the teleologist agree that what

Jones does is morally right because it involves character-

istic (b)."49

Given this, the argument proceeds as follows:

(1) We are free to classify (b) either as a
consequence of the action "lying" or as
Specifying part of the kind of action
Jones does.

(2) If (b) is classified as a consequence,
then what he does is a morally right
action because it has "good consequence
(b) . u

(3) If (b) is classified as specifying a kind
of action, then what he does is a morally
right action because it falls under appli-
cable "rule (b)," where (b) specifies the
content of the rule.

(4) Therefore, we are free to appeal either to
"good consequence (b)" or to "rule (b)" to
justify the rightness of what Jones does.

(5) How we classify (b) is of no moral sig-
nificance.

(6) Therefore it is of no moral significance
whether we justify what Jones does by
appeal to rules or instead by appeal to
consequences.5O

We must admit that the argument is valid. If it is sound
as well, then we must concede with Oldenquist that the
traditional distinction between deontology and teleology
is not one which is well-founded, that in fact "each is

trivial." But has he succeeded in such short order in

141

ringing the death knell of such a venerable and apparently
useful distinction in ethical theory? I would like to argue
that he has not; that, in fact, the distinction is as viable
as it ever was, although the basis upon which the distinc-
tion is drawn may require some careful rethinking.

First, however, one should not fail to be impressed
by the force and initial plausibility of Oldenquist's claim.
If we aaa free to redescribe an act in terms of its conse-
quence(s)--if in Kant‘s illustration we can redescribe X
(telling a lie) in terms of its consequence Y (saving a
life) simply as ‘doing Y‘ (viz. "saving a life‘)--then it
is easy to argue that the controversy between deontologists
and teleologists is at bottom spurious. Even a formalist
like Kant, presumably, would not hesitate to countenance a
rule like ‘One ought to save lives‘ (with appropriate qual-
ifications); in fact, this could even be construed as one
of the "imperfect duties" in Kantian ethics. We would, it

seems, have a prima facie duty to undertake reasonable mea-

 

sures to save lives when it is in our power to do so; at

the very least, saving lives is not prima facie wrong. And

 

one could apply the same analysis mutatis mutandis to any

 

other test case--hypothetica1 or real--which apparently

142

draws the battle line between deontologists and teleolo-

gists: In any case where a prima facie wrong act, X, has

 

consequence(s), Y, whose goodness is sufficient to outweigh
whatever harm (or "evil") may be occasioned by the doing of
X, we may simply redescribe that act as ‘doing Y‘; we can
then formulate the maxim or rule Of which ‘doing Y‘ is an
instance, at which point the deontologist and consequential-
ist can both join hands and commend the agent for doing what
he did. We would then have succeeded in showing that the
deontologist and teleologist, although unbeknownst to them,
were really in agreement all along, and that their blindness
was due simply to their failure to recognize a trivial
metaethical distinction for what it is. We might then
perhaps search for a psychological hypothesis to explain

why men like Kant and Ross, with their peculiar temperaments,
were so attracted to the positions which they espoused, and
why persons like Bentham and Mill, on the other hand, found
that pill so difficult to swallow. At any rate, the so-
called "conflict" between deontologists and teleologists
would no longer be situated in the domain of philosophy

proper.

143

In spite of the initial plausibility of Oldenquist's
position and the validity Of his argument in support Of it,
it should be clear to anyone who has followed my argument in
this chapter that I am forced to disagree with his claims
concerning the relation Of act to consequence, and the bear-
ing of this relation on the metaethical issue of deontology
vs. teleology; it should be equally clear why I disagree.

Let me begin by focusing on premise (l) of Olden-
quist‘s argument: "We are free to classify (b) either as
a consequence of the action ‘lying' or as specifying part
of the kind of action Jones does."51 This "freedom of
classification," as we have already observed, is a function
of what Oldenquist refers to as the "reciprocal translata-
bility of action descriptions and result descriptions."
With respect to this feature of act-descriptions he tells
us:

Our immediate claim is that most of the charac-
teristics of interest to us in such a sequence
[i.e. a sequence consisting of an action and
the consequences of that action] can be de-
scribed by either action descriptions or result
descriptions. If "telling a lie" is the action
description we find convenient, then we are
free to classify the characteristic "a life
being saved" as a consequence; but if we employ
the action description "saving a life," then

only characteristics in the situation which
presuppose saving a life can be described by

144

result descriptions, for example, "grief being

prevented." There is, we are suggesting, no

rigid distinction between action and conse-

quence, and we are free to construe many char-

acteristics either as results or as constitu-

tive of an expanded version of the action.52
This passage rings a familiar bell; Oldenquist‘s character-
ization of the "reciprocal translatability of action de-
scriptions and result descriptions" in substance amounts to
no more and no less than Feinberg's "accordion effect"
thesis as it applies to acts and consequences.

From this it should be apparent on what grounds I
would attack Oldenquist‘s claims concerning the triviality
of the deontological-teleological distinction. Premise (l)
of his argument is too strong as it stands: we are not
(always) free to classify characteristic (b) either as a
consequence or as Specifying part of the kind of action
performed. Our freedom to redescribe acts in terms of
their consequence(s) is not unlimited. In particular, our
freedom in this respect is limited to those cases in which
such redescription is correct or, what amounts to the same
thing, not misleading or infelicitous. And it will be
precisely those sorts of cases which were ruled out as

incorrect in our discussion of Feinberg‘s "accordion effect"

thesis which will also be ruled out in respect to

145

Oldenquist's claims concerning the "reciprocal translata-
bility of action descriptions and result descriptions."

Take his specific example of Jones‘s telling a lie with

the result that a life is saved. To describe this "sequence
of act and consequence" simply as 'Jones saved a life' has
been ruled out on our analysis as incorrect since it is an
instance of a case which violates condition C1 prohibiting

the elision of a prima facie wrong act-type into a descrip-

 

tion of an act which is prima facie right or commendable.

 

Of course, the infelicity could be avoided by expanding or
qualifying the original description in an apprOpriate way;

instead of describing the act as 'telling a lie‘ simpliciter

 

we could, for instance, describe it instead as ‘telling a
life-saving lie.‘ But this is a different matter alto-
gether.

If what I have said is correct, then it would seem
that the traditional distinction between deontological and
teleological ethical theories is not trivial after all.
But out discussion does dramatically demonstrate the crit-
ical importance of selecting correct act-descriptions, and
the care which must be exercised in doing the selecting.

Austin's insistence on the importance of "stating explicitly

146

ghap is being excused" applies with equal force to the
issue at hand, namely, stating explicitly ahap action is
being morally evaluated. With respect to Oldenquist‘s ex-
ample, we have seen that we are app free to redescribe the
act in question simply as 'Jones saved a life'-~period.

We must include reference to the prima facie wrong act of

 

lying in order to avoid infelicity, that is, in order to
forestall any unwarranted inferences or assumptions On the
part of our hearer(s); the descriptionrmnH2either be ex—
panded (e.g. 'Jones told a lie and thereby saved a life,‘
‘Jones saved a life by telling a lie‘), qualified in some
appropriate way (e.g. 'Jones told a life-saving lie'), or
so on. Once we have done this, once we have stated expli-
citly ahap is being evaluated, I think that we will find
that the old metaethical dispute between deontologists and
teleologists re-emerges with renewed vigor. Bentham and
Mill will still hold that the act was good and that the
agent is to be commended for performing it, while Kant
could still remain obstinate, arguing perhaps that he
doesn‘t care how beneficial the consequences may be, lying
is still wrong, or that life-saving lies are nevertheless
still lies and as such are morally wrong, and that therefore

Jones is not to be commended for doing what he did.

147

In summary, it has been argued that Oldenquist's
claim concerning the vacuity of the distinction between
deontological and teleological ethical theories is false,
resting as it does on a mistaken premise concerning the
redescription of acts in terms of their consequences;
namely, his premise asserting the "reciprocal translata-
bility of action descriptions and result descriptions,"
which we have seen is essentially a restatement of Fein-
berg‘s "accordion effect" thesis as it applies to acts and
consequences. Finally, I have tried to Show that once we
take care to state explicitly ahap act is being subjected
to moral scrutiny (i.e. once we have succeeded in selecting
a correct description of the act in question), then we see
that the traditional metaethical distinction between deon-
tology and teleology is as viable as ever. In short, while
Oldenquist has not succeeded in demonstrating that the
deontological-tsleological issue is a moot point, his dis-
cussion does go a long way toward indicating the enormous
importance--practical as well as theoretical--of the prob-

lem of act-descriptions.

148
Notes on Chapter II

1. Redescribing Acts in Terms
of Consequences

1Perhaps you are interested not in what I did (viz.
closed the door), but in hay I did it (by exerting pressure
on it with my hand as Opposed, say, to kicking it shut with
my foot). But note that this would seem to be an appro-
priate response not to the question ‘What did you do?‘ but
rather to something like ‘How did you do it?,‘ a question
which itself seems to presuppose that you already know what
I did.

2Eric D‘Arcy, Human Acts: An Essay in their Moral
Evaluation (Oxford: Clarendon Press, 1963), pp. 15-16.

3Joel Feinberg, "Action and Responsibility," re-
printed in Alan R. White (ed.), The Philosophy of Action
(Oxford: Oxford University Press, 1968), p. 106.

4Ruth Macklin, "Actions, Consequences, and Ethical
Theory," Journal of Value Inquiay, Vol. I, No. 1 (Spring,
1967)] p- 76.

5John E. Atwell, "The Accordion Effect Thesis,"
Philosophical Quarterly, Vol. 19, No. 77 (October, 1969),
p. 337.

 

 

6Feinberg, p. 107.
7Atwell, pp. 337-338.
8Atwell, p. 339.

9Without entering into an extended discussion of
Hart‘s claims concerning defeasibility, I think that this
sort of example constitutes a telling criticism of his
claim that the notion of an action--like certain legal
concepts such as 'contract‘--is a defeasible concept. If
I sign my name to a legal contract under duress, the
presence of the defeating condition (duress) defeats the

149

claim that I have entered into a valid contract; but it
does not defeat the claim that I signed my name to a con-
tract. Similarly, even if Smith is forced to break the
window as a result of Jones's threats, he nevertheless did
break the window. The claim that he (Smith) performed that
action (i.e. breaking the window) is not defeated, even
though the presence of threats may be a mitigating circum-
stance and affect our willingness to ascribe responsibility
and blame to Smith. This analysis is dependent, of course,
on my previous contention that an action sentence such as
‘Smith broke the window,‘ whatever else it does, does in
fact describe an action performed, and does not serve simply
to ascribe responsibility to the agent for the commission
of the act (see Chapter I, section 1).

10Of course, the same analysis holds for the other
two cases of this type. Though Jenkins qaa subordinate
logically cannot promote his peers, he could at a later
time become a superior with the power or authority to hire,
fire, and promote others. And Baker, though he is not a
municipal Official and thus is not in a position or does
not have the authority to approve proposals, could be or
become such an Official with this sort of authority.

llRuth Macklin, "Actions, Consequences, and Ethical
Theory," Journal of Value Inquiry, Vol. I, NO. 1 (Spring,
1967), p. 78.

lef course, I may still be held responsible for my
wife‘s suicide; the fact that my act(s) cannot be correctly
redescribed as 'committing my wife‘s suicide' does pap en-
tail that I am not (at least partly) responsible for her
untimely death. (See Chapter I, section 1, pp. 8-10).
"What this shows," according to Macklin, "is that a person
pap be blamed for consequences of his action which cannot
plausibly be counted as descriptions of his action."
Ibid., p. 77.

13Ruth Macklin, "A Rejoinder," Journal of Value In-
quiry, Vol. I, No. 2 (Summer, 1967), p. 135.

l4Smart‘s illustration is recorded by D‘Arcy in his
Human Acts: An Essay in their Moral Evaluation, pp. 2-3.

 

 

 

 

150

15D‘Arcy, p. 3.

16Feinberg, "Action and Responsibility," p. 106
(italics mine).

17If it is included, then I think the infelicity is
avoided.

l8Feinberg, pp. 106-107.

9Unless, of course, startling-~unlike Opening the
door--is not a "relatively simple act." But even this will
not extricate Feinberg, for in his discussion of simple and
complex acts he treats ‘Open' as a "relatively complex act"
(p. 106).

20D‘Arcy, p. 10.

21D'Arcy, p. 10.

22D'Arcy, p. 12.

23D'Arcy, p. 15.

24D‘Arcy, p. 16.

25D‘Arcy, p. 16.

26D‘Arcy, p. 17.

27D‘Arcy, p. 18.

28D‘Arcy, p. 18.

29D'Arcy, p. 38.

30Oddly enough, it seems to me that the term ‘mayhem'
is not even, properly speaking, an act-term at all and does
not have any sort of moral judgment "built into" it. On
the contrary, 'mayhem‘ more so than any of the other terms
mentioned by D‘Arcy, seems a likely candidate for a term
which is "purely factual and descriptive."

31D‘Arcy, p. 21.

151

32D'Arcy, p. 19.

33D'Arcy, p. 19.

34Joseph Margolis, Values and Conduct (Oxford:
Clarendon Press, 1971), p. 137.

3SMargolis, p. 136.

 

36Margolis, p. 140 (italics mine).

37D'Arcy, p. 39.

38Atwell, "The Accordion Effect Thesis," p. 339.

39Atwell, pp. 339-40 (italics mine).

40Atwell, p. 340 (italics mine).

41We will see shortly that D‘Arcy suggests another
reason why we must rule this description out as incorrect.
However, his reasons for doing so do not, as he supposes,
follow from his statement of the elidability criterion.

42D'Arcy, pp. 19-20.

4
3D'Arcy, p. 32.

44D‘Arcy, p. 33.

45The fact that we found it necessary to add the
riders concerning intentions and anticipatibility of con-
sequences to conditions C2 and C3 but app to Cl suggests
that there may be something to be said for the notion of
"negative utilitarianism." Our unwillingness to allow in
any circumstances the complete elision of prima facie wrong
X into prima facie right or neutral Y suggests an asymmetry
between acts which are productive of pleasure, happiness,
or well-being, on the one hand, and those productive of
pain, misery, or suffering, on the other. The presence of
prima facie wrong X always necessitates some reference to
it if a description is to count as correct. For classical
utilitarianism, given that X produces a given amount of
pleasure and that Y prevents an equal amount of pain, there

 

 

 

152

are no (moral) grounds for preferring the performance of
one over the other. But our analysis of act-descriptions
thus far suggests that act-types which are prima facie
wrong are in some sense Of "greater significance" to human
concerns than those acts which we regard as prima facie
right. (Most laws and moral rules proscribe rather than
prescribe certain types of behavior.) This suggests per-
haps that, all things being equal, one has a greater Obli-
gation to prevent pain or misery than to produce pleasure
or happiness, or at least a greater Obligation to refrain
from producing such pain or misery. Kant, of course, in-
corporated this insight in his distinction between perfect
and imperfect duties.

 

 

2. Implications for Ethical Theory

46Andrew Oldenquist, "Choosing, Deciding, and Doing,"
The Encyclopedia of Philosophy, ed. Paul Edwards (New York:
Macmillan, 1967), Vol. 2, pp. 101-102.

47Oldenquist, p. 102.

48Oldenquist, p. 102 (italics mine).

49Oldenquist, p. 102.

50Oldenquist, p. 102.

51Remember that although Oldenquist uses ‘telling a
lie‘ and 'saving a life‘ in this case as substitution in-
stances Of characteristics (a) and (b), respectively, his
argument is intended to cover all such cases, and is not to
be interpreted to applying to this one case only. Other-
wise his claims could not be construed as general claims
which he obviously intends them to be, as evidenced by his
discussion of the relation between act and consequence, as
well as his actual explicit formulation of the conclusion
(6) of his argument.

52Oldenquist, p. 101.

53In another context, Oldenquist does treat this as
an alternative description of this act (p. 101).

CHAPTER III
AGENT KNOWLEDGE, INTENTION, AND

ACT-DESCRIPTIONS

We have already found it necessary to appeal in
some cases to the agent's intention in acting in order to
determine the correct descriptions of certain types of
actions. In this chapter, I would like to focus more care-
fully on the role Of intention with respect to fixing on
true and correct act-descriptions. The danger here is that
in entering into an inquiry concerning the nature of inten-
tionality, there is the very real possibility of becoming
enmeshed in a plethora of problems, all of which could not
be given justice in a single volume. Among the problems
which one could address in this context are: for an agent
to have done X, is it sufficient that he intended to do X,
or is it merely necessary--or might there be cases in which
it is neither necessary nor sufficient; when and under what
circumstances is the agent‘s intention (or lack of inten-

tion) the decisive factor for selecting the correct

153

154

description of his behavior; is the presence of intention

a constitutive feature of some act-types, or is it merely
incidental to the performance of any given action? These
are only a few of the highly complex and difficult issues
which could be dealt with in a discussion of intentionality.
Rather than attempting to deal with each of these questions
satisfactorily--a task well beyond the SCOpe of this study--
I would instead like to examine one specific issue in detail
and try to Offer an adequate account of this one specific
feature of action. The question I propose to deal with is
this: what is the relation between intentional action, on
the one hand, and an agent‘s knowledge or awareness of what
he is doing, on the other; and, finally, what is the bear-

ing of this relation on the problem of act-descriptions?

1. Knowing What One is Doing

 

In a given set of circumstances, as we have noted,
doing X may constitute doing Y; lying, when done under oath
in a court of law, just la to commit perjury. This in it-
self is neither surprising nor problematic. However, once

we take into account an agent‘s intention in doing this or

155

that, we are faced with a couple of interesting questions.
Consider: if an agent intends to do X while he is X-ing,
and X-ing constitutes Y-ing (or X-ing entails Y-ing) in
those circumstances, does it follow that the agent thereby
intends to do Y? If, for example, a man intentionally
raises his arm at an auction (and given that raising one‘s
arm at an auction = making a bid), does he thereby intend
to make a bid? The answer in this case is obviously no;
intentionally raising one's arm at an auction does not
thereby entail that the agent intended to make a bid. And
this is so for a very simple reason--namely, the agent may
not know (or believe) that raising one‘s arm in this con-
text = making a bid. In other words, if an agent is ig—
norant of the fact that X-ing constitutes (or entails)
Y-ing in a given set of circumstances, then it is not only
conceivable but quite obvious that he could intend to do X
without thereby intending to do Y. In this respect, ‘in-
tending‘ is analagous to other intentional verbs--'knowingfl
‘believing,' ‘doubting,' ‘desiring,' and so on. One can
know (doubt, believe) that Jones is over six feet tall
without thereby knowing (doubting, believing) that the

president of the bank is over six feet tall (even though

156

Jones may happen to be the bank president); likewise, one
can desire X without thereby desiring Y which happens to

be identical with X under a different description. This
feature Of intention is not surprising or controversial,
and is not, as far as I can tell, contested by any of those
dealing with the issue of intentional action.

Having said this, we are now in a position to form-
ulate a more intriguing question: granted that an agent
need not intend to do Y in intending to do X (where X and Y
are different descriptions Of what the agent is doing) due
to ignorance on his part, what about the case in which not
only does X, in fact, constitute Y, but the agent knows or
believes this as well? That is, if in a given set of cir-
cumstances X = Y, and in addition the agent knows or be-
lieves that X = Y, can he then intend to do X without
thereby intending to do Y? For example, say Jones goes
to an auction not as a spectator at some ritual which is
strange to him, but as an active participant; he has been
to auctions before, is fully acquainted with the rules and
procedures, knows how to make bids, abstain from making
bids, and so on. Now among the things which Jones knows

is the following: when one raises one's arm at an auction,

157

he is making a bid on the merchandise on the block at a

given moment; that is, he knows that in these circumstances

 

raising one‘s arm = making a bid. Now let us ask whether,
given this state of affairs, Jones could intend to raise
his arm without thereby intending to make a bid. Unlike
the case considered above in which an agent due to ignor—
ance could intend to do X without thereby intending to do
Y (where X = Y), there is no consensus among philOSOphers
concerning how to answer the question now before us.

Some philOSOphers, notably G. E. M. Anscombe, Opt
for an affirmative answer to our question; that is, an agent
may intend to do X without thereby intending to do Y (when
X = Y) even if the agent knows or believes that X = Y.
Anscombe considers this possibility in her now-famous ex-
ample of the man pumping water.1 We are to suppose a man
pumping water into a cistern thereby replenishing the
supply of drinking water in a nearby house. Furthermore,
the water which the man is pumping has been systematically
contaminated with a deadly cumulative poison whose effects
are unnoticeable until they can no longer be cured. The
man who has contaminated the source (who is not identical

with the man doing the pumping) has done so with a View to

158

assassinating the party chiefs inhabiting the house, who
are engaged in some sinister plot which the would-be
assassin hopes to abort. Now let us suppose that the man
doing the pumping knows that the water has been poisoned
and the purpose for which it has been poisoned, even though
he is not the one who contaminated it. Now we may ask:

can this man (the pumper), knowing full well that the water
supply has been poisoned, intend to replenish the supply of
drinking water to the house (X) without thereby intending
to poison its inhabitants (Y)? Anscombe‘s answer is an
unequivocal yes:

A further difficulty however arises from the
fact that the man‘s intention might not be to
poison them but only to earn his pay. That is
to say, if he is being improbably confidential
and is asked 'Why did you replenish the house
water-supply with poisoned water?,' his reply
is, not ‘TO polish them Off,‘ but ‘I didn't
care about that, I wanted my pay and just did
my usual job.‘ In that case, although he knows
concerning an intentional act of his--for it,
namely replenishing the house water-supply, is
intentional by our criteria--that it is also

an act of replenishing the house water-supply
with poisoned water, it would be incorrect, by
our criteria, to say that his act of replenish-
ing the house supply with poisoned water was
intentional. And I do not doubt the correct-
ness of the conclusion; it seems to Show that
our criteria are rather good.2

159

I for my part do doubt the correctness of this conclusion;
and the dubiousness of this conclusion seems to me to show
that Anscombe's criteria are, on the contrary, rather bad.
Now it might be objected that this passage from
Anscombe does not commit her one way or the other on the
question of whether 'poisoning the inhabitants' was an in-
tentional action on the part Of the agent. All she is
claiming is that even though, in the given circumstances,
'replenishing the house water-supply‘ = 'replenishing the
house water-supply with poisoned water,‘ only the former
is an intentional action on the part of the agent; and
this in itself implies nothing concerning whether ‘poison—
ing the inhabitants‘ is or is not an intentional action on
the part of the pumper. To this I make two replies. First,
the initial sentence of the passage quoted strongly sug-
gests that Anscombe herself sees her argument as showing
that poisoning the inhabitants is not an action the pumper
intended to perform. Secondly, even if this passage itself
does not commit Anscombe to the claim that the man did not
intend to poison the inhabitants, nevertheless the argument
contained therein does. For, if his replenishing of the

house water-supply succeeded in poisoning the inhabitants,

160

we could then ask, 'Why did you poison those people?‘ And
again he could answer, 'I didn‘t care about that, I wanted
my pay and just did my usual job.‘ And this again would,
presumably, show that poisoning the inhabitants was not an
intentional act on his part, just as this same answer to
the question 'Why did you replenish the house water-supply
with poisoned water?‘ was sufficient for showing that this
action was not intentional on his part. (This interpreta-
tion of Anscombe‘s position will be further substantiated
in the next paragraph.) But finally, one might Object,
even if what I have said is correct, I have suddenly shifted
the context of discussion. For now I am no longer talking
about one action being the same as another (like replenish—
ing the house water-supply in Anscombels case = replenishing
the house water-supply with poisoned water), but instead I
am now talking about the consequence of some action on the
part of the agent. For poisoning the inhabitants is not
the same as replenishing their house water-supply with
poisoned water, but at best a certain and foreseeable con-
sequence of doing the latter. And this I admit. But since
I intend all my claims in this chapter as well as the

analysis I will subsequently suggest to apply to cases like

161

phaaa as well as to cases where doing X, strictly speaking,
constitutes doing Y, there will be no need to qualify my
remarks when talking about cases in which Y is the fore—
seeable consequence of X. Therefore, from this point on-
ward I will use the formula ‘X = Y' to mean either (a) in
the given circumstances, X in fact, is the same as or con-
stitutes Y (as raising one‘s arm at an auction = making a
bid), or (b) in the given circumstances, Y is the certain
and foreseeable consequence of X (so that, in Anscombe‘s
case, replenishing the house water-supply with poisoned
water = poisoning the inhabitants). The context of dis-
cussion in each instance will make clear which sense of
‘X = Y' is intended. With these provisos in mind, then,
let us return to our original question, namely, if X = Y,
and if A knows or believes that X = Y, can A intend to do
X without 'doing Y' thereby qualifying as an intentional
action on A‘s part? We have already seen that Anscombe
replies affirmatively.

Before proceeding to argue for what I take to be
the correct answer to our question, I would first like to
point out what sorts of considerations are not relevant to

deciding this issue, and thereby forestall certain ways of

162

attempting to substantiate an affirmative answer to our
question in the manner Of Anscombe. What I want to suggest
in this context is that the notion of wanting or desiring
is not relevant--or at least not relevant in the way that
Professor Anscombe and others think--to the solution of

the problem as here stated. If this is ignored, it is easy
to see how one may be tempted by an affirmative answer to
our question. For if we suppose that one cannot intend
that which he in no wise wants or desires, then it is clear
that Anscombe's pumper does not intend the poisoning Of the
inhabitants Of the house in replenishing the house water-
supply, even though it seems that he is quite willing to
poison them by doing this, or that our man at the auction
may not intend to make a bid in raising his arm. For, by
hypothesis, ‘poisoning the inhabitants‘ and 'making a bid‘
are not things that our respective agents were aiming at

or positively wanted or desired to bring about. Thus, if

a necessary condition of intending to do Y is that the
agent must want, desire, or aim to bring it about that Y,
then it follows that if Y (‘poisoning the inhabitants‘ and
‘making a bid,‘ respectively) is not something which the

agent wanted, disired, or aimed at in doing what he did,

r We 'e.‘h}“"

163

then Y cannot count as an intentional action on his part
(even though he did intend to X--the same act under a dif-
ferent description). Anscombe does seem to be assuming
something like this in her discussion of the notion of a
"desirability characterization" and its relationship to
intentional action. She says in one place that "desira-
bility characterizations are required in the end for pur-
posive action."3 In other words, there must be some fea-
ture in the act intended which the agent finds desirable,
something which he (rightly or wrongly) believes to be good
or worthwhile, either intrinsically or extrinsically. From
this it would follow that if an act X did not possess such
a desirability characterization, then X could not qualify
as an intentional action with respect to a given agent.

And as I have been contending, it is precisely this assump-
tion which lends credence to the view that the pumper, for
instance, need not have intended to poison the inhabitants
of the house in replenishing the house water-supply with
poisoned water even though he knew or believed that, in the
given circumstances, 'replenishing the house water-supply

with poisoned water' = ‘poisoning the inhabitants.‘ On

the other hand, if this assumption itself is incorrect,

 

164

we may have grounds for rejecting Anscombe's answer and for
establishing, on the contrary, that the pumper gag intend
to poison the inhabitants as well as replenish the house
water-supply or, alternatively, that ‘poisoning the in-
habitants‘ was as much an intentional action on the part

of the pumper as was ‘replenishing the water-supply.‘

What, then, can be said for this assumption; namely,
that a necessary condition for an act X to qualify as in-
tentional is that the agent desire or want to bring it
about that X, or that X must possess some relevant "desira-
bility characterization?" The first thing that can be said
about this claim is that Anscombe is surely not alone in
endorsing it. Fleming, for example, says that "to do a
intentionally is to do a knowing you are doing it, wanting
to do it, and doing it because you want to do it."4 For
this reason, he concludes with Anscombe that "if a person
doing a does not want to do y . . . then, although in doing
a he is doing y, he is not doing y intentionally."5 In his

Lectures on Jurisprudence, Austin defines an intention to

 

do something as follows:

a present intention to do a future action may
be defined to be: "A present desire of an
Object (either as an end or as a means),
coupled with a present belief that we shall

165

do acts hereafter for the purpose Of attain-
ing the Object.6

Returning to more recent times, Kenny states that "an agent
intends an action only if he (a) knows he is doing it and
(b) wants to do it either for its own sake or in order to
do some other action."7 Churchland:

One may say, ‘X A-ed because he wanted ¢,'
but one might also say, 'X A-ed with the
intention of bringing about O.‘ Now I
should claim that the explanatory force of
these two statements is precisely the same,
and indeed that these two statements are
materially equivalent.

 

Audi claims that
it seems . . . plausible to hold that in-
tending to do A . . . entails being dis-
posed tO trying to do A if certain Obstacles
should arise; and [this] would support the
view that intending entails wanting.
In the next paragraph, Audi concedes that even if these

considerations about intending and trying do not prove

conclusively that intending entails wanting, they never-

 

theless do, when taken together with other considerations,

"make it quite plausible to hold that intending entails

 

wanting."10 Finally, George Pitcher, in arguing against
those who claim that if A intends to do X, knowing that

X = Y, then he thereby intends to do Y, claims that "we

166

can describe many . . . examples where our linguistic intui-
. . . . . 1
tions would be outraged if we adhered to this prinCiple." 1
On the other side of the issue, there are those who
wish to dispute both that (a) intending entails wanting, and
(b) the claim founded on (a), namely, that one can intend to
do X without thereby intending to do Y, even where X = Y and
the agent knows or believes this to be the case, because the
12 13

agent may not want to do Y. Both Aune and Bedford take
this position in rebuttal to Kenny. Bedford cites what he
takes to be a counter-example to Kenny's claim that intend-
ing entails wanting:

For there is plainly a sense in which inten-

tional action does not entail wanting to do

the action ‘for its own sake or to do some

other action.‘ I do not want to go to the

dentist, but when I walk to his surgery I do

so with the intention of keeping my appoint-

ment.l4
Notice here that Bedford uses the phrase "intentional ac-
tion" rather than "intending to," and denies that inten-
tional action entails wanting; this distinction will figure
importantly later. Although at times his position appears
ambiguous, Chisholm seems to come down on the side Of those

who Oppose both (a) and (b) above:

Unless we use ‘desires to do it,‘ as it is
sometimes used, to mean merely having the

167

intention Of doing it, this addition will not

change matters. For a man may have the in-

tention of doing what he has no desire to do.15
In addition, Chisholm‘s "principle Of the diffusiveness Of
intention" seems to bolster this interpretation; this he
states as follows:

I would say that, if a rational man acts with

the intention of bringing about a certain state

of affairs p and if he believes that by bring-

ing about p he will bring about the conjunctive

state of affairs, p and q, then he does act

with the intention of bringing aboup the con-

junctive state of affairs, p and q. 6
Finally, Castaneda comes down strongly on the side of those
who claim that intending does not entail wanting.17

Although I have classified these philosophers with

respect to their stance on the issue Of whether intending
entails wanting, I do not wish to imply that there is com-
plete agreement within each camp; on the contrary, there is
as much in-fighting among the members of a given camp as
there is disagreement between the camps themselves. I have
listed these positions taken by various philosophers simply
to Show that the controversy is not new (and therefore not
one that I am inventing); secondly, it demonstrates that,

in attacking the principle that intending entails wanting,

although I may be in the minority, at least I do not

168

constitute a minority of one (if that were the case, my
confidence would be considerably shaken). I do not intend
to rehearse all the arguments and counter—arguments proposed
by these philOSOphers. Instead, I will suggest an argument
independent of those employed by any of the authors cited,
an argument which, I feel, is sufficient for showing that
wanting is app a necessary condition for intentional action.
Before proceeding, I would like to make a couple Of
Observations concerning the tenor and overall progress Of
the debate between the philosophers just cited. In many
instances, the argument (and counter-arguments) proceed
somewhat as follows: first the defender Of the claim that
intending entails wanting proposes this thesis, if not as
a self-evident proposition, at least as a principle which
no reasonable person could or would wish to dispute. The
Opponent then Suggests one or more counter-examples de-
signed to Show that an agent can intend to do what he in
no way desires or wants. The defender then insists that,
nevertheless, there is or must be some sense of ‘want‘ in
which it is still true to say that the agent wants what he
intends. And very Often the end result of the debate is

something like the following: the defender keeps insisting

169

that if it is true that the agent intends to do this or
that, then he must in some sense want to do it; in this
case, as Chisholm points out, the person seems to be guilty
of using "‘desires to do it‘ . . . to mean merely having

the intention of doing it,"18 which simply seems to beg the
question. Or else, in the process of arguing that the agent
in the apparent counter-examples "really did want (in some
sense) to do what he intended to do," the concept Of 'want-
ing‘ becomes so stretched beyond anything with which we are
familiar that the thesis then becomes uninteresting.

Be that as it may, it is now time to return to the
problem at hand. I want to argue against Anscombe that
wanting to do X is not a necessary condition for intending
to do X. Now I do not, of course, want to deny what Fleming
refers to as "the closeness of the connection between want-

ing and doing intentionally."19

I am the first to admit
that this connection obtains for many--indeed most--in-
stances of intentional action. After all, why would one
aim at or intend to perform a given action if there were
not some feature in the action itself (or its anticipatible

consequences) which the agent finds it desirable to bring

about? Nevertheless, as a general claim supposed to be

170

applicable to all instances Of intentional action, this

assumption does not seem, at least prima facie, without

 

exceptions. And this sort of case under consideration la
peculiar, a case in which doing X in a given set of circum-
stances = doing Y and, furthermore, the agent knows or be-
lieves this to be the case. It seems clear to me that in
any such case we can only conclude that Y (the act under
a description in which it is not aimed at or desired) is
just as intentional as X (the same act under a different
description aimed at by the agent, or the description under
which the act does possess the required desirability char-
acterization). As applied to our illustrations, then,
‘poisoning the inhabitants‘ and ‘making a bid' are as much
intentional actions on the part of the agents in the given
circumstances as are ‘replenishing the house water-supply‘
and 'raising one‘s arm,‘ respectively, even though one
member Of each of these pairs lacks a desirability char-
acterization on Anscombe's analysis.

So far all I have done is indicated that my intui-
tions dictate an answer to our question which is at Odds
with that which Anscombe gives. However, we are not forced

to Simply give it up at this point and write it off as a

171

clash of intuitions and therefore an insoluble problem.
Rather, let us examine the consequences of the affirmative
answer to our question to determine whether such conse-
quences are or are not acceptable; if they are not, then
it would seem reasonable to reject Anscombe's solution on
independent grounds.

Let me begin by supposing that Anscombe‘s solution
does, in fact, lead to counterintuitive consequences. That
is, I will suppose for a moment that it runs counter to our
enlightened intuitions to claim that Y-ing is not an inten-
tional action on A‘s part, even though A's act of (inten-
tionally) X-ing = Y-ing and, furthermore, that A knows or
believes this to be so. One way in which to defend a
theory which has counterintuitive consequences is to argue
that the merits of the given theory so far outweigh such
consequences (particularly if they are not momentous), that
one ought to forego one‘s intuitions and accept the theory
as it stands. Non—Euclidean geometries, for example, have
several counterintuitive consequences; in spite of this
feature, however, the merits of such systems in terms of
what they enable us to do are taken as compelling grounds

for their adoption. This is not, however, the line of

172

defense which Anscombe takes. On the contrary, she feels
that her position is in no need of any defense whatsoever;
she takes what I have been regarding as the counterintui-
tive consequence of her theory of intentional action as
according with our intuitions and thereby providing addi-
tional evidence to confirm her theory of intentional action.
However, I would like to argue that not only is this conse-
quence counterintuitive, but that if we do accept it we
thereby commit ourselves to other claims which are clearly
unacceptable, thus providing independent grounds for re—
jecting Anscombe's answer.

Consider, for example, the case of an act-type

which is prima facie wrong, say killing. If A kills B,

 

we will (or ought to) reserve a precise moral judgment of
A's act and A‘s character until such time as all the rele-
vant facts and circumstances surrounding the killing are
made known to us. We will typically want to know such
things as whether it was done "in the heat of passion,"

or whether it was coolly and deliberately calculated,
whether the action was undertaken voluntarily or whether
the agent was coerced, and so on; our final moral judgment

of the act and the agent will hinge on such facts as these.

173

And among the more important Of these facts is whether the
killer disposed of his victim intentionally or not. For to
say that an act is intentional often involves the ascription
Of responsibility and praise or blame. In short, we behave
differently toward an agent whose act is intentional than we
do toward the agent if what he did was not intentional. In
particular, all things being equal, we will be inclined to
judge the killing and the killer less harshly if the killing
were done unintentionally (or non-intentionally) than if it
were an intentional act on the part of the agent. In other
words, absence of intention on the part of the agent, al-
though it is not always an excusing circumstance (with re-
spect to wrong acts), is the sort of thing which functions

as at least a mitigating circumstance in most cases. We

 

may not retract our verdict that Jones is to blame for
Smith‘s death simply because he sincerely avows that he
didn't mean (i.e. intend) to do it (we may hold him culpable
in virtue of negligence, for example), but surely there will
be a difference between our judgment of Jones in this case
and our judgment of the Mafioso hit man who intentionally
does away with his victim. To establish that Jones‘s kill-

ing Smith was not intentional-~though in itself not a

174

complete justification (as, say, a successful plea Of self-
defense would be)--nevertheless is a circumstance which
counts against the initial presumption of guilt and inclines
us to view the offense more sympathetically or leniently.
Let us now apply these remarks to Anscombe‘s illus-
tration. Let us examine whether, in the case under consi—
deration, the pumper‘s sincere avowal that he didn‘t even
care about poisoning the inhabitants (he just wanted to
earn his pay and so was doing his usual job, viz. pumping
water) would be sufficient to establish that 'poisoning the
inhabitants‘ was not an intentional act on his part. Ac-
cording to Anscombe‘s analysis it does; it seems to me,
on the contrary, that it does not even hagla to Show that
‘poisoning the inhabitants' was not an intentional act on
the part of the pumper. For if it did suffice to show that
this act was not intentional then it could also, according
to the analysis I have suggested, to some extent at least
"let the agent Off the hook" for committing that act, or
incline us to view the episode more leniently. But it seems
clear to me that we would not countenance the pumper's
sincere disclaimer that he wanted to poison the inhabitants

in any way as a mitigating (or much less an exculpating)

175

circumstance, as we would countenance, say, a candid admis-

sion of ignorance on his part that, in these circumstances,

 

‘pumping the water' = ‘poisoning the inhabitants.‘

If the reader is not convinced by these observa-
tions, let me suggest a derivative argument. If one grants
my major premise, viz. that if a reason R is sufficient for
showing that a given act is not intentional, then R is also
capable of functioning to identify a relevant mitigating
circumstance, think Of the consequences for law and moral-
ity on Anscombe‘s analysis. A defendant in a court of law,
for example, might be able to hOpe for a reduced sentence
or a lesser charge (if not an acquittal) if he were only
able to convince the judge and jury that, even though he
knew full well that, in the circumstances under considera-
tion, killing his rival = increasing his own chances for a
promotion, it was really only the important promotion-—and
not his rival's death--that he wanted; or, in other words,
that 'getting the promotion,‘ unlike ‘killing his rival,‘
possesses the necessary desirability characterization.
After all, to the prosecutor‘s question, 'Why did you kill
your rival?‘ he might sincerely reply, 'Oh, I didn‘t care

about that; I wanted the promotion desperately and just did

176

what I had to do.‘ But this is patently absurd. On the
basis of all this, I conclude that the notion of a desira-
bility characterization is a frequent--but not necessary--
condition of intentional action. And since it is only the
assumption that this notion is a necessary condition which
lends plausibility to the view that A can intend to do X
but nevertheless not intend to do Y, even though he knows
(or believes) that X = Y, I therefore reject Anscombe‘s
analysis and her answer to our question. On the contrary,
the man at the auction who knows that ‘raising his arm‘ =
‘making a bid‘ has intentionally made a bid in raising his
arm (even if all he wanted to do was scratch his armpit);
similarly, Anscombe‘s pumper who knows that, in the given
circumstances, ‘replenishing the water-supply‘ = ‘poison-
ing the inhabitants‘ has intentionally poisoned the in-
habitants in replenishing the water-supply (even though
all he really wanted was to earn his wagesff"In other
words, if Y is not only the foreseeable result of doing X
in a given set of circumstances, but foreseen by the agent
himself as well, then if doing X is to count as an inten-

tional action on his part, then so is doing Y.

177

Before proceeding, I would like at this time to meet
a possible objection to the account offered here. One might
argue that absence of intention on the part of the agent
does not function as a mitigating circumstance in the kinds
of cases under consideration because the agent is neverthe-
less willing to bring about the untoward consequence of his
act. Thus, one might insist, this is why we don‘t or won‘t
let the agent off the hook in such cases. For even though,
for example, the pumper was willing to poison the inhabi-
tants or was willing to bring about their deaths, this
nevertheless is not what he intended. Or alternatively,
one might argue that although absence of intention on the
part of the agent in such cases does function as a mitigat—
ing circumstance, nevertheless his willingness to perform
this (wrong) action nullifies those grounds for mitigation.
My reply to this sort of argument is that, with respect to
such cases, the only grounds for claiming that although A
was willing to do X, his doing X nevertheless is not in-
tentional on his part, is that doing X is not something
which A really wanted. But this is the very point being
disputed. There is certainly a sense in which a person may

be willing to see a certain undesirable state of affairs

178

brought about without thereby intending that. Thus, for
example, I may be willing to see the United States become

a second-rate power without thereby intending to do any-
thing to bring this about. But the sense of ‘being willing
to‘ here is more akin to acquiescing. In such cases, one
may be 'willing to . . .‘ without thereby intending any-

thing, precisely because the person is not acting. But the

 

sense of ‘is willing to‘ with which we are concerned is not
like this at all. On the contrary, in the sorts of cases
under consideration, the agent is not simply willing to let
certain things happen--he is willing tg_dg these things, or
willing to do something else of which these things are cer-
tain and foreseeable consequences. Thus, Anscombe‘s pumper
is not only willing to see the inhabitants die, but he is
willing to poison them. Similarly, the defendant in our
court case is not only acquiescing in the death of his
rival, but he is willing to kill him to get the promotion.
And in these sorts of cases, I can see no grounds for dis-
tinguishing ‘A was willing to do X‘ from ‘X was intentional
on A‘s part‘ (unless, of course, one falls back on the
thesis that intending entails wanting). If we know that

A was willing to X and he then went on to do X, we will

179

certainly hold him responsible and praise or blame him for
X-ing; and it is precisely these same features which make
X-ing intentional on A‘s part.

I have been careful for the most part throughout my
argument to avoid the locution 'A intended to do Y,‘ opting
instead for 'Y was an intentional action on A‘s part.‘
Thus, if X = Y and A knows or believes this, then if A in-
tended to do X, Y is also thereby an intentional action on
the part of A. There is a sense in which the distinction
between these two sorts of locutions is "merely verbal."
For I am also prepared to say of the agent in such cases
that he did, in fact, intend to do Y in intending to do X.
But in another sense the distinction does capture one im-
portant insight of those who wish to claim that intending
does entail wanting. There is a sense of ‘intending‘--
though not the only sense--in which to say that A intended
to do X suggests that X i§_something that A wanted, either
as an end in itself or as a means to something else which
he desires; although it might be more accurate to refer to
this as a species of intending, rather than as a distinct
sense of the word ‘intend.‘ The idea behind this is that

in those cases in which doing X entails doing Y (and the

180

agent knows or believes this as well), even though the
agent may only want to do X, he has no choice about Y. If
one insists, we could restrict the locution ‘A intended to
. . .‘ in such cases to X, although I like Castaneda and
Chisholm see no necessity for doing 30.20 But what is
important to note is that whether or not we adopt this
convention has no bearing whatsoever on the question of
whether the "unwanted action" is or is not intentional.
Nevertheless, most writers who, like myself, argue that
wanting is not a necessary condition for intending do dis-
tinguish between these two species of intending, namely,
that which is and that which is not accompanied by the
corresponding want or desire on the part of the agent.
This is precisely what Bentham had in mind in distinguishing
direct from oblique intentions. Jack Meiland captures this
insight by distinguishing purposive from non-purposive in-
tentions,21 where, roughly, purposive intentions to X are
those in which the agent wants to do X and a non-purposive
intention has reference to an action which the agent has
no particular desire to perform (or even, perhaps, does
not want to perform). With respect to this issue, I am in

complete agreement with Meiland‘s account:

181

If a purposive intention is inseparable from

a non-purposive intention [i.e. if wanting

to do X entails doing Y], then the agent has

the non-purposive intention solely because

he has the purposive intention.22
And Castaneda, for his part, distinguishes "primary" from
"derivative" intentions:

Primary are the intendings an agent has by

themselves so to speak; derivative are those

intendings which he has only because he has

both other intendings and certain believings

which he takes to imply the former intend-

ings. This is all as it should be. Intend-

ing is like believing, and one has primary

as well as derivative beliefs (i.e. believ-

ings).23
Once again, the important thing to notice is that no matter
what type of intending is involved in a particular case--
direct or oblique, purposive or non-purposive, primary or
derivative--they are all intendings, and the actions thus
intended are intentional. And it is precisely for this
reason, namely, to preserve the distinction between these
two species of intending, that I was careful to distin-
guish ‘A intends to do X‘ from ‘Y is an intentional act on
A‘s part‘ in our previous discussion. As Audi remarks,
"it is very natural to assume that intentional actions are

those brought about by intentions."24 But, as he goes on

to show, this is not always the case. An action may count

182

as intentional on the part of the agent even though he
never explicitly formulated the intention to perform that
action prior to performing it. Intentional actions are the
sorts of activities for which we are held responsible and
praised or blamed, and for this it is not necessary, as I
have been trying to show, that the action in question is
something which the agent wanted to do. I agree here with
Audi when he says that "the relation between the notion of
intending and that of intentional action is not so close
as one might well think, though the paradigms of intentional
action are those performed with some further intention."25
Finally, let me emphasize that my countenancing a
distinction between these two species of intending in no
way obviates the dispute between myself and Anscombe. One
might be tempted to think that we are arguing at cross-
purposes, since my argument is designed to show that doing
Y (which, though not wanted by the agent, is entailed by
his doing X) is an ‘intentional action‘ on the part of the
agent, whereas what Anscombe is denying is only that, in
such cases, the agent ‘intended to do Y‘ (where 'intended
to‘ is understood in the sense in which it is accompanied

by wanting). But this is not so, for Anscombe herself

183

frames her argument to prove that "it would be incorrect

. . . to say that his act of replenishing the house water-
supply with poisoned water was intentional."26 In summary,
then, we are at loggerheads; I am denying what Anscombe
asserts, namely, that the act in question was not inten-
tional on the part of the pumper.

My argument thus far in this section has been
rather lengthy and, at times I fear, somewhat convoluted.
Let me attempt to outline the major steps of the argument
before proceeding to discuss its bearing on the problem of
act-descriptions. We began our present inquiry by asking
whether it is possible for an agent to do X intentionally
(in a set of circumstances in which doing X = doing Y),
without thereby intending to do Y, when the agent himself
knows (or believes) that doing X = doing Y. We saw that
Anscombe defends an affirmative answer to this question; I,
on the contrary, wished to argue for a negative answer.

The general structure of my argument may be outlined as
follows: let A be Anscombe‘s pumper, let X = ‘replenishing
the house water-supply with poisoned water,‘ and let Y =
'poisoning the inhabitants‘; let us assume further that the

pumper knows that, in these circumstances, X (‘replenishing

184

the house water-supply with poisoned water‘) = Y (‘poison-

ing the inhabitants‘).

1)

2)

3)

On Anscombe‘s view, A‘s sincere denial
that he was aiming at or desired to bring
it about that Y is sufficient for dis-
qualifying Y as an intentional action on
A‘s part, since Y then lacks a necessary
"desirability characterization" with re-
spect to A. (Remember that the pumper's
reply to the question ‘Why did you re-
plenish the house water-supply with poi-
soned water?‘ was ‘I didn‘t care about
that, I wanted my pay and just did my
usual job.‘).

But any feature which is cited as a reason
for holding that a given action on the
part of an agent is not intentional ipso
facto is capable of functioning also to
identify a mitigating (although not neces-
sarily an exculpating) circumstance.

But A's sincere denial that he wanted or
was aiming at Y (poisoning the inhabitants)
in doing X would never be countenanced as
a mitigating or exculpatory circumstance
if, by hypothesis, A knew full well that
doing X = doing Y; his denial, then, would
not be sufficient to establish that Y-ing
(poisoning the inhabitants of the house)
was not an intentional act on A‘s part.

It should now be clear that the upshot of my argu-

ment against Anscombe is this: knowledge on the part of

the agent that doing X = doing Y is sufficient for quali-

fying doing Y as an intentional action on his part if

doing X counts as intentional on his part. Here I fully

185

agree with Gustafson's observation that "ignorance or lack
of awareness are the root notions in unintentionality;
their Opposites are the root notions in intentionality,

whether intentional activity or intentional objectivity."27

2. Bearing on Act-Descriptions

 

All of this may, perhaps, seem quite unexceptional.
However, when we begin to examine the bearing of the agent‘s
knowledge or ignorance of what he is doing on the problem
of correct act-descriptions, the picture becomes a little
more complicated. Once we have drawn together the implica-
tions of various claims I have defended throughout the
course of this inquiry, it might appear to some that I
have backed myself into a corner. What I want to do now
is to describe this corner and then try to show that, even
if I have backed myself into it, it is nevertheless not
such an ugly place to be after all.

It will be recalled that during my examination of
Rayfield‘s analysis of the notion of a correct act-
description (Chapter I, section 3), I was critical of his

claim that a necessary condition for the correctness of an

186

act-description of the form ‘A did X' is that A must know
that he is (was) X-ing while he is (was) X—ing. It is
important to repeat that I did not at that time reject this
claim outright as false. Rather, my reason for rejecting
Rayfield‘s analysis is that it rested on certain assump-
tions and resulted in various consequences which I deemed
unacceptable. Specifically, his position rested on the
assumption that an agent could be praised or blamed for
an action only if the agent himself knew that he was per-
forming the action in question. What I want to do even-
tually is to indicate what I take to be right in Rayfield‘s
analysis and suggest an alternative account which will
capture his insights while avoiding what I consider the
unacceptable results and assumptions.

I have just argued in the previous section that
knowledge on the part of the agent with respect to what
he is doing is, to use Gustafson‘s phrase, the "root notion"
in intentionality. Now, it might be urged, in a case in
which doing X = doing Y, if (i) A did, in fact, knowingly
and intentionally do X, and furthermore (ii) A knows that
doing X = doing Y in the given circumstances, such that

(iii) doing Y was also an intentional action on A‘s part,

187

then is this not sufficient to qualify the sentence ‘A did
Y‘ or ‘A Y-ed‘ as a correct description of A's action?

The answer to this question is, of course, ‘Yes,
unless . . . ,‘ where the ellipsis may be filled in by one
or more of our defeating conditions. So that, in the case
of the southern sheriff, for example, in which killing one
innocent Negro entails saving four lives, although (i) it
is true that he knowingly and intentionally killed the
innocent Negro, and moreover (ii) he knew that, in the
given circumstances, killing the innocent Negro = saving
four lives, such that (iii) saving four lives was also an
intentional act on the part of the sheriff, nevertheless
to describe what he did simply as ‘saving four lives‘ would
be incorrect, since it violates condition Cl governing the
elision of act-terms into consequence-terms. In the case
of Anscombe‘s pumper, on the other hand, where replenish-
ing the water-supply with poisoned water = poisoning the
inhabitants, given that (i) he did, in fact, knowingly and
intentionally replenish the house water-supply with poi-
soned water, and (ii) he knew (or had every reason to sus—
pect) that this = poisoning the inhabitants, such that

(iii) poisoning the inhabitants also counts as an

188

intentional action on the pumper's part, we may correctly
describe what the pumper did as ‘poisoning the inhabitants
of the house,‘ since this does not violate any of our con-
ditions for eliding act-terms into consequence-terms.
After all, poisoning the inhabitants of a house is, on
general grounds, an anticipatible consequence of replenish-
ing their water-supply with poisoned water. The important
point to notice about this illustration is that I do not
regard the fact that poisoning the inhabitants was not
something that the pumper positively wanted or desired as
a condition that would defeat the correctness of the de—
scription, ‘The pumper poisoned the inhabitants of the
house.‘ (Of course, I do not mean to imply that this is
Anscombe's position; she is not addressing herself to this
issue at all.) In other words, we need not-~indeed ought
not--append to the list of three conditions above a fourth
condition specifying that X is something that A, in fact,
wanted or desired (i.e. that the action in question have
the necessary desirability characterization with respect
to A).

Finally, just what are we to say concerning the

illustration of the man at the auction who is unfamiliar

189

with the conventions and procedures? Let us begin with
conditions (ii) and (iii). In this case, as I have argued,
since A did not know that he was Y-ing (where Y = making a

bid) even though he knew he was X—ing (where X = raising

 

his arm) and intended to do X, making a bid would not count
as an intentional action on his part. This is similar to
the case of the foreigner driving in America who extends
his arm from the car (X) with the intention of pointing out
a church to his companion; in this case, we are to suppose
that the driver does not know that X-ing (extending one‘s
arm from an automobile while driving) = Y-ing (signalling

a turn). In both of these cases, assuming that A did X
intentionally (raised his arm/extended his arm), but is

not cognizant that doing X = doing Y in the given circum-
stances (i.e. making a bid and signalling a turn, respec-
tively), we would expect a candid, affirmative answer to
the question, ‘Did you X?‘ (i.e. ‘Did you raise your arm?‘
and ‘Did you extend your arm?‘), but a candid, negative
response to the question ‘And did you Y as well?‘ (i.e.
'Did you make a bid?‘ and ‘Did you signal a turn?‘). Now,
apart from disqualifying doing Y in both instances as an

intentional action on the part of the agent, what else

190

would be the force of this sincere disclaimer on the agentks
part? In particular, what bearing would it have on the
question of whether the agents in our illustrations did,

in fact, make a bid and signal a turn? In other words, the
question we must now deal with is ‘Qid A Y?‘; that is,did
the man make a bid, and did the foreigner actually signal a
turn? Is it true to say of them that they made a bid and
signalled a turn, respectively? If so, would it be correct
as well to describe their actions as ‘making a bid' and
‘signalling a turn‘?

I must confess that I am not quite certain how to
handle cases of this sort. On the one hand, I am inclined
to think that it would be true to say of our agents that
they had made a bid and signalled a turn, but that never—
theless the act-descriptions, ‘He made a bid' and ‘He sig—
nalled a turn‘ must be counted as incorrect. The problem,
however, is this. My reasons for thinking that these de-
scriptions would not be correct would appear, on the face
of it at least, to count also against their being true.

Let me attempt to explain what I have in mind here.
Let us focus for a moment on the man at the auc-

tion. It seems to me that to describe what he did simply

191

as ‘making a bid‘--without any qualification or expansion--
would suggest or "imply" that this was something done in-
tentionally on his part. And since, by hypothesis, the man
did not know that raising his arm = making a bid, and
therefore did not intend to make a bid, this description
would be misleading. Yet, one might plausibly argue, mak—
ing a bid is precisely that kind of activity which must be
undertaken intentionally by an agent; that, in other words,
there is no such thing as making a bid unintentionally,
just as there is no such thing as committing murder un-
intentionally (in one of the most common senses of
‘murder‘).28 And if this is so, then my reason for think-
ing that the act—description ‘He made a bid‘ is incorrect
in the weak sense of being misleading would also function
to show that the very same description is indeed incorrect
in the strong sense of being false. For, if intentional-
ity is "built into" the act-term ‘bidding‘ as it is with
respect to act-terms like ‘murder,‘ then to say of our

man that 'He made a bid‘ would be not only misleading but
strictly false. Still, I am inclined to believe that this
description, while misleading, is nevertheless true. Is
there any way of extricating myself from this difficulty

and justifying this belief?

192

(i) Intentionality
and Act-Terms

 

 

What I would like to do is introduce and exploit a
distinction similar to that introduced by Prof. Henry
Leonard with respect to various types of attributives.29
He distinguishes existence-entailing attributives such as
‘hit,‘ ‘hang,‘ etc. which entail the existence of some ob-
ject (in both ‘A hit B‘ and 'A hanged B‘ there must be some
object, B, which is the object of the hitting and hanging,
respectively), from attributives which entail non-existence,
for example, ‘. . . is fictitious‘ (to say that ‘Pickwick
is a fictitious character‘ entails that there is no such
object as Pickwick). Finally, he distinguishes both of
these kinds of attributives from those which entail neither
the existence nor the non-existence of their corresponding
"objects"; these attributives are, so to speak, "neutral"
with respect to existence. An example of such an attribu-
tive would be ‘is afraid of‘; Jones may be just as afraid
of ghosts as he is of his mother-in-law who, in fact, lives
next door.

Similarly, with respect to act-terms, I would like

to distinguish those which entail intentionality on the

193

part of the agent from those which entail neither the pres-
ence or absence of intention on his part. 'Murder,‘ for
example, would be a paradigm case of an act—term which does
entail the presence of intention on the part of the agent
(in the sense of ‘murder‘ specified above). As mentioned
previously in another context, it is impossible to commit
murder unintentionally; conversely, to say that ‘A mur-
dered B intentionally‘ would be redundant. Most act-terms,
however, entail neither the presence nor the absence of
intention on the part of the agent. A can kill, hit,
nudge, offend, or mislead B either intentionally or unin-
tentionally. In other words, most act-terms, unlike
‘murder,‘ do not have intentionality "built into" them

as one of their specifying characteristics. They are,
instead, "neutral" with respect to the circumstance of
intentionality.

Finally, we must determine whether there is a class
of act-terms which entail the absence of intention on the
part of the agent analagoustjmaLeonard‘s class of attri—
butives which entail the non-existence of their correspond-
ing "intentional objects." Certain candidates for such

act-terms suggest themselves immediately--'manslaughter‘

194

for instance. But on reflection, I think a case could be
made for holding that the term ‘manslaughter‘ is not even
an act-term at all, much less an act-term having certain
features. To say that an agent committed an act of man-
slaughter dgeg entail that he did something without inten—
tion on his part. But precisely what is this "something"
which he did unintentionally? Is it not the case that what
he did was to kiii some other person unintentionally?
‘Killing‘ is the act-term in question; ‘manslaughter‘ serves
to qualify this act-term, to inform us what type of killing
it is that the agent committed. Although this argument
does seem to have at least some initial plausibility, I am
afraid that it simply won‘t do. For one might take the
very same reasons I have just advanced for the view that
the term ‘manslaughter‘ is not an act-term and force me

to "testify against myself," by showing that these same
reasons entail that ‘murder‘ is not an act—term either.

One could argue as follows: To say that an agent committed
murder does entail that he did something intentionally.

But precisely what is this "something"? Is it not the case
that what he did was to kiii another person intentionally?

‘Killing‘ is the act—term in question; ‘murder‘ serves

195

simply to qualify this act-term by informing us what type
of killing it is that the agent committed. Therefore,
either both ‘murder‘ and ‘manslaughter‘ refer to species
of killing and thus are act-terms in their own right; or
else both ‘murder‘ and ‘manslaughter' function merely as
qualifiers, informing us what type of killing has been com-
mitted (where ‘killing‘ is the act-term in question).
Nevertheless, perhaps it is significant that there is no
verb corresponding to the substantive ‘manslaughter.’ One
can commit an act of manslaughter just as one can perform
an act of mercy; but one cannot "manslaughter" someone any
more than one can "mercy" someone. ‘Murder,’ on the con-
trary, can function either as a verb or as a noun. But I
certainly would not want to base much on these grammatical
observations; perhaps they are significant, but on the
other hand, perhaps they are not.

Fortunately, however, I think there is a much
stronger argument, independent of any of the foregoing
observations, that one can employ to show that any such
class of act-terms which would entail the absence of in—
tention on the part of the agent must necessarily contain

no members. I must agree with Rayfield's contention that

196

for any item of behavior X to qualify as a candidate for an
action of any kind--bound or unbound, intentional or unin-
tentional--it is a necessary condition that X be such that
someone could on some occasion decide to or intend to do X.
Similarly, any term which denotes a type of action must
refer to a species of activity which it is possible to
undertake intentionally. A term which purports to describe
some activity which is impossible to decide or intend to
perform would, for this very reason, not count as an ESE“
term at all. It is difficult to conceive how such "act-
terms" could ever be introduced into our vocabulary in the
first place or, if they were, what possible function they
could serve. For these reasons, then, I think we can
safely rule out this class of act-terms, namely, act-terms
which entail the absence of intention on the part of the
agent.

We are left, then, with two mutually exclusive and
jointly exhaustive classes of act-terms; first, there are
those act-terms which entail intentionality on the part of
the agent and, secondly, there are those act—terms which
entail neither the presence nor the absence of intention,

those that are "neutral" with respect to the circumstance

197

of intentionality. Each of these classes of act-terms will,
of course, contain many members (although I think it is
reasonable to suppose that the class which is neutral with
respect to intentionality will contain a far greater number
than the other class). The following are lists of act-
terms typical of each of these classes; needless to say,
these do not purport to be exhaustive:

Act-terms which entail the presence of inten-

 

tion: ‘murder,' ‘lie,‘ ‘steal,‘ 'fast,‘
‘cheat,‘ ‘conspire,‘ ‘embezzle,‘ ‘defraud,‘
etc.

Act-terms which are neutral with respect to
intentionalipy: ‘nod,‘ ‘hit,’ ‘hurt,’
‘offend,‘ ‘kill,‘ ‘wound,‘ 'annoy,‘ etc.

 

Now it is left to us to determine to which class ‘making a
\ . ' I Q ' 30
bid and Signalling a turn belong.
Before proceeding, I should like to examine briefly

a curious claim which Rayfield makes concerning certain
types of act-terms; in particular, regarding act-terms
which characterize various sorts of criminal homicide, he
writes:

no one is ever in a position to describe ter

[his] own behavior as murder, or any other

type of criminal homicide. The reason for

this is that ‘murder,’ 'manslaughter,‘ and

‘negligent homicide‘ (and sometimes ‘causing

suicide‘) are never just terms used to de-
scribe someone‘s behavior. Rather, these

198

are verdicts or charges which can only be
brought within a court of law. A better way
of putting this is that ‘murder,’ ‘man-
slaughter,‘ and ‘negligent homicide‘ are
classifications of items of behavior which
can only be used by a court of law, and then
only if the behavior in question has already
qualified as an action. Hence, ‘Are you
murdering?‘ and ‘Did you murder?‘ are not
questions which can legitimately be used in
determining whether a given item of behavior
qualifies as an action.31

 

Let us consider the act-term ‘murder.‘ (For the
reasons indicated above, I am not even sure whether we
ought to refer to terms like ‘manslaughter‘ and 'negligent
homicide‘ as act-terms at all). Even if we grant that
‘murder' is never a term used iEEE to describe someone‘s
behavior, but is also a verdict or charge brought within a
court of law, it still would not follow that it could not
be used to describe someone's behavior. After all, neither
is ‘donating a large sum to a worthwhile charity‘ iEEE a
phrase used to describe someone's behavior; there is a
commendatory force in this description as well. Neverthe-
less, it would not follow that we could not use the phrase
‘donating a large sum to a worthwhile charity' to describe
what someone did. So in the case of ‘murder,’ it seems to
me that it may be quite intelligible in some cases to say

something like ‘Jones murdered his wife, even though the

199

jury acquitted him‘ (or the grand jury refused to indict,

or even though he was never even a suspect in the case, or

 

whatever). Jones may subsequently confess to the murder--
notice, not just to the killing--of his wife. It seems to
me that Jones may very well know about his own knowledge,
purpose, degree and kind of recklessness, presence or ab-
sence of intention, and so on, and thus be in a good posi-
tion to correctly describe his own action as ‘murder‘ in-
dependently of whether a formal charge is brought or whether
a verdict is rendered by a jury in a court of law.

Returning now to our principal question, how are we
to classify ‘making a bid‘ and ‘signalling a turn?‘ Are
these act-terms or phrases which entail intentionality on
the part of the agent, or are they neutral with respect to
the circumstance of intentionality? In our examination of
Oldenquists' suggested proposal for identifying appropriate
act-descriptions (Chapter I, section 3), we briefly dis-
cussed the role of social practices and conventions in the
determination of act-types. There we remarked that although
the presence of applicable practices or conventions is often
not a necessary condition for an action to be performed,

their presence is nevertheless sometimes sufficient for a

 

200

specified type of action to have been performed--regardless
of whether or not the agent himself intended to perform
that action. And it seems to me that ‘making a bid‘ and
‘signalling a turn‘ are paradigm cases of just this sort of
"convention-governed" action. The man at the auction who
raises his arm in order to scratch his armpit has, in doing
so, made a bid on the merchandise, even though it was not
his intention to do so, indeed even though he did not know
that this act on his part constituted making a bid. And
the relevant practices and conventions dictate that the
foreigner who extends his arm from his automobile has
signalled a turn, in spite of whatever else he intended to
do and in spite of whether he, in fact, knew that he was
signalling a turn. In both cases, it seems to me, the
agent may very well be held accountable for what he did.

If our man at the auction was the last to bid on the mer-
chandise, then he may legitimately be held responsible for
his bid and forced to pay up to meet his obligation. And
should the foreigner‘s signal cause some untoward chaos at
the intersection, he may be hauled into the station and

ticketed for improper lane usage or whatever.

201

Now it is important to emphasize that I am 22E
claiming, nor am I in any way committed to the claim that
ignorance on the part of the agent that his doing X = doing
Y will have no effect on our description of what he did or
on our subsequent behavior toward him. As I mentioned in
my argument against Anscombe, showing that an act was per-
formed unintentionally is the sort of thing which can
function as a mitigating circumstance. One can imagine
the unwitting bidder complaining to the auctioneer that
he didn‘t intend to bid on the merchandise, that he didn‘t
even know that he was doing so; that the auctioneer had,
so to speak, "misinterpreted" what he was doing. And the
auctioneer may be particularly sympathetic with the man‘s
plight, release him from his financial obligation, and
pick up the bidding again at the point where he left off.
The auctioneer might do this, but he need not. Most im-
portant, the auctioneer is gs; guilty of misinterpreting
the unwitting bidder‘s behavior since, in those circum-
stances, the relevant practices and conventions dictate
that raising one‘s arm = making a bid. Likewise, with
respect to the foreign driver, his sincere claim of ig-

norance might be countenanced as a mitigating circumstance.

202

He might be left off this time with a stern warning and a
brief lecture on the rules of the road. But, once again,
he need not be left off; after all, "ignorance of the law
is no excuse." In other words, although A‘s sincere plea
of ignorance that he was doing X may--not must--be coun-
tenanced as a mitigating circumstance in such cases, A‘s
ignorance nevertheless does s93 defeat the claim that he
did, in fact, X.32 If all this is correct, then it would
be £522 to say of the agents in our examples that ‘He made
a bid' and ‘He signalled a turn,‘ respectively. And if it
is true to say these things of them in spite of the absence
of intention on their parts to perform these actions, it
follows that the act—phrases ‘making a bid' and ‘signalling
a turn‘ need not entail the presence of intention on the
part of the agent; these act-phrases, therefore, are prop-
erly classified as instances of act-terms which are neutral
with respect to the circumstance of intentionality.

I have now succeeded, I think, in establishing my
first claim concerning such act-descriptions. When I began
the discussion of these illustrations I confessed that,
although I was not quite certain how to handle cases of

this sort, I was inclined to say that the descriptions

203

‘He made a bid‘ and ‘He signalled a turn‘ were Eggs act-
descriptions even if incorrect in the weak sense of being
misleading. If my argument thus far is sound, I have suc-
ceeded in showing that it is Eggs to describe what our
agents did as ‘making a bid‘ and ‘signalling a turn.‘ What
remains, then, is to (i) show that such descriptions, while
true, are nevertheless misleading and thus incorrect in the
weak sense, and (ii) to offer an analysis to account for

the misleading nature of such descriptions.

(ii) Attributing
Intentionality

 

 

Let us return to our classification of act-terms.
We have thus far distinguished between those act-terms
which entail the presence of intention on the part of the
agent and those which are neutral with respect to inten-
tionality. What I would like to do now is to examine the
"implications" of action-sentences which employ act-terms
from each of these classes. This examination will, I trust,
provide the key to accounting for the misleading nature of
certain types of act-descriptions of which ‘He made a bid‘

and ‘He signalled a turn‘ in our examples are instances.

204

Let us begin with the first class of act-terms,
those which entail the presence of intentionality on the
part of the agent. With respect to any act-term, X, from
this class, an action-sentence of the form ‘A X-ed‘ will,
if true, entail that A, in doing X, was acting intention-
ally. For example, the sentence ‘A murdered B‘ (as 'murder'
is being understood throughout this discussion), if true,
entails that A was acting intentionally in disposing of his
victim. It follows that if A did not kill B intentionally,
then the act-description ‘A murdered B‘ will be incorrect
in the strong sense of being false. The same would hold
true of any description of the form 'A did X‘ or ‘A is
X—ing‘ in which the act-term or act-phrase substituted for
‘X‘ belongs to that class of act-terms which entail the
presence of intention on the part of the agent. In such
cases, what a sentence of the form ‘A did X‘ or 'A is-
X-ing‘ formally entails with respect to the agent's inten-
tions will coincide exactly with what that sentence "im-
plies" or suggests to one‘s audience.33

But is the situation the same for action-sentences

containing act-terms or phrases belonging to the second

class, those which are neutral with respect to

205

intentionality? That is, although such act-terms obviously
do not entail either the presence or absence of intention,
is it true that they "imply" neither the presence or ab—
sence of intention as well? In this instance, I want to
argue that, with respect to intentionality, what an action-
sentence of the form 'A X-ed‘ or ‘A is X-ing‘ entails is
not necessarily the same as what it "implies" or suggests.
Consider a couple of action-sentences containing

act-terms which are neutral with respect to intentionality:

1) Smith hit Jones and wounded him seriously.

2) Adams killed Brown with a slingshot.

What I wish to claim concerning sentences such as these is
that, although the act-terms in question (‘hit,‘ ‘wound,’
‘kill‘) belong to that class which entail neither the pres—
ence nor the absence of intention on the part of the agent,
such locutions--without any qualification or expension--
nevertheless do "imply" or suggest that what was done in
each case was done intentionally. 'In sentence (1), for
example, to say simply that 'Smith hit Jones and wounded
him seriously‘ is to suggest that Smith hit Jones inten-

tionally with the further intention of wounding him; this

206

holds in spite of the fact that hitting and wounding are
things which can be done with or without intention on the
part of the agent or, in other words, in spite of the fact
that ‘hit‘ and ‘wound‘ are act—terms which entail neither
the presence nor the absence of intention on the part of
the person doing the hitting or wounding. And the same
considerations apply to sentence (2).

It may be objected that the examples of act-terms
which I have chosen are particularly well-suited to support
my claim, but that I have been biased in my selection.
Perhaps, one might argue, it is true that action-sentences
of the simple form ‘A did X' or ‘A X-ed‘ gs "imply" that
A X—ed intentionally when act-terms like ‘hit,‘ ‘wound,‘
or ‘kill‘ are substituted for 'X,‘ but that this may not
hold when other act—terms are employed, even when such act-
terms belong to the class which is neutral with respect to
intentionality. Or even stronger, perhaps it is the case
that not only are there act-terms of this sort which do 223
"imply" the presence of intention, but that there are even
cases of such act-terms which, though neutral with respect
to intentionality, nevertheless "imply" the absence of

intention on the part of the agent when they are employed

207

in simple action—sentences of the form ‘A did X.‘ And I
must admit that I certainly have not exhaustively canvassed
all the act-terms belonging to this class. In all honesty,
however, I am not able to think of a plausible candidate
for an act-term in this class having this feature. Maybe
an act-term like ‘offend‘ could be suggested as a likely
candidate. But even here it seems to me that to say simply
that ‘A offended B‘ "implies" that he did so intentionally,
even though one can offend either intentionally or uninten-
tionally. If we suppose that A did, in fact, offend B, but
that he did so unintentionally, this could be captured by
saying something like 'A‘s behavior offended B‘; but notice
that this expression.does not have the form ‘A X-ed.‘ But,
one might object, further, we do use locutions like ‘A

deliberately offended B,‘ and that this would suggest that

 

the act-term ‘offend' "implies" as well as entails neither
the presence nor the absence of intention on A‘s part.
For, if in saying that ‘A offended B‘ it is "implied" that
the offending was intentional, then the locution ‘A is-

liberateiy offended B‘ would be redundant. But it seems

 

to me that the insertion of the word ‘deliberately‘ in

such cases serves to emphasize the speaker's distaste for

208

A‘s behavior, and not to inform us what species of offen-
siveness he is guilty of.

At any rate, what I am trying to argue for here is
not, I think, a feature peculiar to act-terms. A similar
case can be made with respect to Leonard‘s class of attri-
butives which entail neither the existence nor the non-
existence of their corresponding "intentional objects."

As J. R. S. Wilson notes,

On an extensional account . . . a psychological
statement containing a referring expression
would imply that the referring expression had
a referent. Now if I utter a psychological
statement employing a name or definite de-
scription I certainly seem to imply that this
name or description has a referent. Thus if I
say that Smith is angry with the man who stole
his bicycle, or that Jones is afraid of the
burglar who broke in last night, I seem to
imply that someone did steal Smith‘s bicycle,
or that a burglar did break in last night.
Someone who heard me would conclude from what
I said that this was so, and would surely be
justified in doing so. Furthermore, if I
found out that I had been mistaken in think-
ing that there was such a person, I would
probably retract or modify my statement.

In other words, just as expressions containing predicates
which are neutral with respect to existence "imply," though
they do not strictly entail, the existence of the corre-
sponding "intentional object," so too expressions contain-

ing act-terms neutral with respect to intentionality"imply,"

209

though again they do not strictly entail, the presence of
intention on the part of the agent.

If this is correct, then we are finally able to
substantiate our claim concerning the original illustra-
tions, namely, that although ‘He made a bid‘ and ‘He sig-
nalled a turn‘ are true act-descriptions, they are never-
theless incorrect in the sense of being misleading. For
if we were right in classifying ‘making a bid‘ and ‘sig-
nalling a turn‘ as belonging to the class of act-terms
which are neutral with respect to intentionality, then a
sentence like ‘He made a bid,‘ although it does not entail
that the bid was made intentionally, nevertheless "implies"
or suggests that the agent intended to make the bid. And
finally, since we have been supposing throughout that the
agent did BEE intend to make a bid, this act-description
would be misleading and therefore incorrect in the weak
sense. Likewise, in the case of the foreign driver, al—
though it is Eggs to say that ‘He signalled a turn,‘ this
act-description is nevertheless misleading since it sug—
gests or "implies" that he intended to signal the turn.
This analysis also enables us to account for the peculiarity

of the action-sentence ‘He alerted a prowler‘ in Davidson‘s

210

illustration. Although alerting prowlers is something one
can accomplish either intentionally or unintentionally,
nevertheless to say simply that ‘He alerted a prowler' is
to suggest that the agent intended to do so; since, by
hypothesis, this is not the case, this act-description must
be counted as misleading, even though it is true that the
agent did succeed in alerting the prowler.

In each of these cases, of course, the infelicity
could be avoided by expanding or qualifying the given act-
description. We could relate the entire episode in detail,
including all of the relevant circumstances (one of which
would be the absence of intention on the part of the agent).
Alternatively, we could simply qualify the given act-
description with a word or phrase like ‘unintentionally,‘
‘inadvertently,‘ ‘accidentally,’ ‘unwittingly,’ or so on,
and thus "cancel out" the initial presumption that the
action was performed intentionally by the agent. ~Thus,
for example, with respect to the bidder, we might say
something like 'He made a bid, although he didn‘t mean to‘
or ‘He accidentally (or inadvertently) made a bid on the
lamp‘; or with respect to Davidson‘s case we might say that

the man ‘accidentally alerted the prowler,‘ and so on.

211

Although I hope that the point I am making is clear, let me
conclude this discussion by explicitly formulating a final
condition for correct act-descriptions to handle the sorts
of cases we have been considering in the latter portion of
this chapter (i.e. the unwitting bidder and foreign driver):
C6: In general, one may not correctly describe
what an agent did by a simple action-
sentence of the form 'A X-ed,‘ where X
is an act-term neutral with respect to

intentionality, unless the agent knew that
he X-ed and intended to do X.35

Conclusion

Let me conclude the discussion in this chapter by
indicating what I take to be right in Rayfield's original
analysis of the notion of a correct act—description, as
well as pointing out our areas of disagreement.

First for the disagreements. I do not consider
Rayfield‘s conditions for correct act-descriptions suffi-
cient, though I do regard each of them as necessary. It
will be recalled that on his analysis, a description Dn is
correct if and only if: (i) Dn is true, and (ii) Dn is

the same as Dn‘, where Dn‘ is the description of A‘s act

212

under which A X-ed. Condition (ii) simply amounts to
knowledge or awareness on the part of the agent that he
did, in fact, X. I regard both (i) and (ii) as necessary,
but not jointly sufficient, for qualifying a given descrip-
tion of A‘s act X as correct. In the case of the southern
sheriff, he did save four lives (i.e. the description ‘he
saved four lives' is true) and, furthermore, he knows that
he saved four lives. Nevertheless, on my analysis, ‘He
saved four lives‘ does not qualify as a correct act-
description since it violates Cl. In general, the correct-
ness of a given act—description meeting both of Rayfield‘s
conditions may nevertheless be defeated by violating one
or more of the conditions Cl-C6.

On the other hand, I do agree with Rayfield in
regarding the agent‘s awareness or knowledge of what he is

doing or has done under a given description as a necessary

 

condition for the correctness of that description. Con-
versely, A‘s ignorance that he was X—ing is sufficient to
disqualify 'A X-ed‘ as a correct description of A‘s action.
What is important to notice, however, is that I do not
adopt this view for the same reason as Rayfield. On the

contrary, I reject Rayfield‘s argument that failing to

213

acknowledge A‘s ignorance as a sufficient condition for
disqualifying ‘A X-ed‘ as a correct description has the
unacceptable consequence that one might be praised or
blamed for doing X unwittingly. Against this I argued
that, although we would not praise an agent for "doing
good" unwittingly, we nevertheless can and do in some cases
hold agents responsible and blame them for doing X unwit-
tingly; in other words, I reject Rayfield‘s assumption that
the conditions for ascribing praise and blame are symmet-
rical. I also went on to argue against Anscombe‘s claim
that the absence of the necessary desirability character-
ization is sufficient for disqualifying an item of behavior
as an intentional action on the part of an agent. Against
this I argued that her notion of a desirability character-
ization is not a necessary condition for, although it
almost invariably accompanies, intentional action. Rather,
I regard ignorance on the part of the agent that he is
(was) doing X sufficient for disqualifying ‘A is X-ing‘
or ‘A X-ed‘ as a correct act-description for the following
reasons:

Case (i): In the sentence schema ‘A X-ed,‘

where X is an act-term which entails

the presence of intention on the
part of the agent, if A is not aware

Case (ii):

214

that he is X-ing then X does not
qualify as an intentional action
on A‘s part; therefore, the de-
scription ‘A X-ed‘ in such cases
is incorrect in the strong sense
of being false.

If, on the other hand, the act-term
substituted for ‘X' in the simple
sentence schema ‘A X-ed‘ belongs to
the class of those neutral with
respect to intentionality, then
although the description 'A X-ed‘
may be true, it is nevertheless
misleading if the agent did not
know that he was X-ing, since it
"implies" or suggests--though it
does not entail--that A intended

to do X.

215

Notes on Chapter III

 

1. Knowing What One is Doing

1 .
G. E. M. Anscombe, Intention, 2nd. ed. (Ithaca:

Cornell University Press, 1957), p. 37.

 

2Anscombe, pp. 41—42.

3Anscombe, p. 75.

4Brice Noel Fleming, "On Intention," The Philosoph—
ical Review, Vol. LXIII, No. 3 (July, 1964), p. 307 (italics
mine).

 

 

5Fleming, p. 307.

6John Austin, Lectures on Jurisprudence (London:
John Murray, 1873), Vol. I, p. 451.

7Anthony Kenny, "Intention and Purpose," The Journal
of Philospphy, Vol. LXIII, No. 20 (October 27, 1966),Eh 647.

 

 

8Paul M. Churchland, "The Logical Character of
Action-Explanations," The Philosophical Review, Vol. LXXIX,
No. 2 (April, 1970), pp. 230-231.

 

9Robert Audi, "Intending," The Journal of Philosophy,
Vol. Lxx, No. 19 (July 19, 1973), pp. 390—391.

10Audi, p. 391 (italics mine). As I will indicate
later, Audi adds certain qualifications which make his posi-
tion come much closer than might be expected to the one I
wish to defend. In fact, in one place in this same article,
Audi insists that his remarks do 22E commit him to "denying
the validity of the inference from (a) x intends to do A,
(b) x believes that A = A1,and (c) x is rational, to (d) x
intends to do A1" (Ibid., p. 397, n. 10).

 

1George Pitcher, "‘In Intending‘ and its Side Ef-
fects," The Journal of Philospphy, Vol. LXVII, No. 19 (Oc-
tober 8, 1970), p. 664.

 

216

2Bruce Aune, "Intention and Foresight," The Journal
of Philosophy, Vol. LXIII, No. 20 (October 27, 1966),
pp. 652-654.

l3Erroll Bedford, "Intention and Law," The Journal
of Philosophy, Vol. LXIII, No. 20 (October 27, 1966),
pp. 654-656.

l4Bedford, p. 654.

 

 

 

 

15Roderick Chisholm, "The Structure of Intention,"
The Journal of Philosophy, Vol. LXVII, No. 19 (October 8,
1970), p. 645 (italics mine).

l6Chisholm, p. 636. The ambiguity in Chisholm‘s
position that I mentioned above is generated by another of
his principles that would seem to conflict with the prin-
ciple just stated. This second principle he refers to as
"the principle of the nondivisiveness of intention": "from
the fact that (i) there is a certain state of affairs p
such that a man acts with the intention of bringing it
about that p occurs, and (ii) that state of affairs p en-
tails or includes a certain other state of affairs q, and
furthermore (iii) the man in question knows or believes
that p entails g and (iv) he is completely rational, it
does not follow that (v) he acts with the intention of
bringing it about that g occurs" (Ibid., p. 636). I think,
however, that the inconsistency between these two principles
is only apparent. For the principle of the diffusiveness
of intention entails that in a case of this sort, the agent
acts with the intention of bringing it about that both p
and g occur; but this is quite compatible with the prin—
ciple of the nondivisiveness of intention which entails
that in a case of this sort it does not necessarily follow
that the agent acts with the intention of bringing it about
that 3 alone occurs. This interpretation is borne out by
Chisholm's analysis of test cases. For example, Chisholm
asks: "If a men acts with the intention of bringing it
about that he will visit Paris and if he knows that General
de Gaulle will be there, does he act with the intention of
bringing it about that state of affairs which is his being‘
in Paris while General de Gaulle is there?" and he answers,
"It seems to me that the answer is yes" (Ibid., p. 636).
In other words, step (v) of the principle of the

 

217

nondivisiveness of intention is not to be read as ‘bringing
it about that q occurs is psi an intentional act on the
part of the agent,‘ but rather that ‘bringing it about that
g alone occurs is not the whole of the agent's intention;
what is intentional on his part is the bringing it about
that both p and g occur.

7Hector-Neri Castaneda, "Intentions and the Struc-
ture of Intending," The Journal of Philosophy, Vol. LXVIII,
No. 15 (August 5, 1971), pp. 453-466. See also Annette C.
Baier, "Act and Intent," The Journal of PhilOSOphy, Vol.
LXVIII, No. 19 (October 8, 1970), pp. 648-658.

l8Chisholm, p. 645.

19Fleming, p. 309.

20See Castaneda, p. 459.

21Jack W. Meiland, The Nature of Intention (London:

Methuen & Co. Ltd., 1970), pp. 7-13.

22Meiland, p. 14.

23Castaneda, pp. 459-460.

24Audi, p. 401.

25Audi, p. 401.

26Anscombe, pp. 41-42.

7Donald Gustafson, "The Range of Intentions,"
Inquiry, Vol. 18, No. 1 (Spring, 1975), p. 89.

2. Bearing on Act—Descriptions

28The O.E.D. defines the verb ‘murder‘ as ‘to kill

(a human being) unlawfully with malice aforethought‘ and
defines ‘malice‘ as ‘wrongful intention generally.‘ This,
of course, is what is referred to in legal terminology as
first degree murder. It is this sense of ‘murder‘ which

218

I have in mind here and which I will be employing throughout
the subsequent discussion.

9Henry S. Leonard, "Essences, Attributes, and Predi-
cates," Presidential Address to the Western Division of the
American Philosophical Association, printed in the Proceed-
ings and Addresses of the American Philosophical Association,
Vol. XXXVII (1963-64).

30Of course, a simple test could be devised for as-
certaining to which of these classes a given act-term be-
longs. Given any act-term, X, if it makes sense to say
‘A did X, but he did so unintentionally,‘ then the act-term
in question will belong to the class which is neutral with
respect to intentionality; if it does not make sense, then
the act-term will belong to the class of those which entail
intentionality. Unfortunately, the present sorts of cases
are precisely those in which it is not clear how we should
answer this question. Thus the need for a more careful
examination.

31Rayfield, "Cody‘s Categories," p. 427.

32Notice that it is A‘s plea of ignorance in such
cases that may be counted as a mitigating circumstance,
and psi his disclaimer that X was, in fact, something that
he positively wanted or desired or, in other words, his
disclaimer that X possessed the necessary desirability
characterization. If the bidder did, in fact, know that
raising one‘s arm at an auction = making a bid, then his
sincere avowal that "all he wanted to do was scratch his
armpit" would in no wise be regarded as a mitigating or
exculpating circumstance. Given the relevant practices
and conventions, he did make a bid; furthermore, given
his knowledge that raising one‘s arm at an auction =
making a bid, making a bid was an intentional action on
his part.

33Notice that I restrict the coincidence of what is
formally entailed and what is "implied" or suggested in
such cases to the circumstance of intentionality. It could
still happen that a sentence of the form ‘A X-ed‘ (where
‘X‘ stands for an act-term entailing intentionality), while
having the same entailments and "implications" with respect

219

to the presence of intention on the agent‘s part, could
still be misleading for independent reasons (it might, for
example, violate one of the conditions specified in Chap-
ter II). In such a case, of course, the "implications"
and formal entailments of the sentence 'A X-ed‘ will not
coincide exactly.

34J. R. S. Wilson, Emotion and Object (Cambridge:
Cambridge University Press, 1972), p. 146.

 

5We must append one other "unless" clause to this
condition. For it can happen that the results dictated by
C6 may sometimes conflict with the conditions which include
reference to the anticipatibility of consequences particu-
larly C3. In the case of the man who talks to the bus
driver and (unintentionally) distracts him, C3 permits us
to say simply that ‘He distracted the bus driver,‘ in spite
of the fact that this was not the intention with which he
talked to the driver, since distracting drivers is, on
general grounds, an anticipatible consequence of talking to
them while they are driving. But when we do encounter a
conflict of this sort, it seems that we must let C3 take
precedence. For, at least in the context of morality, it
is more important that we avoid the one kind of infelicity
than the other. That is, we will insist in cases covered
by C3, that the act be described as ‘doing Y' even if it
was not intentional on the agent‘s part, since he should
have known that Y was the natural consequence of his X-ing.

 

 

CHAPTER IV

CONCLUDING REMARKS

What I have attempted to do in this thesis is,
first, to precisely formulate an issue which, though it
heretofore has not received much direct attention, is not
only of intrinsic philosophical interest, but also, as I
have tried to show, of extreme practical importance as
well. This I have referred to as the problem of act-
descriptions, and it can be stated as follows: given that
any item of human behavior can be truly described in an
infinite number of ways, are there any criteria for choos-
ing from this class that description which is the "pre-
ferred" or correct description of a given action? I indi-
cated the importance of this enterprise for such areas as
the philosophy of mind and ethics. Convincing the reader
that the "problem of act-descriptions" does constitute a

bona fide philosophical problem the solution to which has

 

an important bearing on practical issues was my initial

aim in writing this thesis; this I trust I have achieved.

220

221

Having stated the problem, the remainder of the
study was devoted to solving it, that is, to arriving at
a satisfactory analysis of the notion of a ‘correct‘ act-
description, an analysis which is at once both useful and
illuminating while preserving (or at least not seriously
violating) our pre-analytic intuitions concerning human
acts and their appropriate characterizations. In this
context I first examined, and ultimately rejected, the
only explicit attempts to date to solve the problem of
act-descriptions as I have formulated it here. In their
place, I substituted an analysis of a ‘correct‘ act-
description which, although perhaps not as rigorous or
precise as those rejected, nevertheless avoids the diffi-
culties engendered by the very rigor and precision of those
alternative analyses; an analysis which, moreover, is more
"faithful to the facts" and which does perform the job
required of it.

The foregoing in effect constitutes a synopsis of
my goals and achievements in Chapter I. And this is
perhaps the most important part of the entire inquiry.
For what is provided therein is a general "formula" or

guideline, however inexact, for testing various

222

descriptions of a given action for correctness. The rest
is simply a matter of applying this formula to specific
cases. Of course I am not implying here that such appli-
cations will be purely mechanical; surely it will not. Nor
am I suggesting that there will never be disagreement at
any point during the application. But having furnished the
general guidelines, it is my contention that when specific
disagreements do surface, there is nevertheless a general

framework within which reasonable parties will be able to

 

test their disagreement and usually reach a consensus. In
this respect, it is not an overstatement to say that the
rest of the thesis stands or falls with the success or
failure of my analysis.

In Chapters II and III I put my analysis to work
on the problems of the relation between acts and conse-
quences and acts and intentions, and the bearing of these
relations on the issue of act-description. In Chapter II,
I began by asking when and under what circumstances we may
correctly redescribe an agent‘s act solely in terms of its
consequences. First, I demonstrated that some such re-
descriptions are simply false and thus incorrect in the

strong sense. I then proceeded to the more interesting

223

problem, namely, that of distinguishing those true rede-
scriptions of acts in terms of their consequences which
count as correct as well from those which do not, according
to the analysis adopted in Chapter I. In discussing a
number of types of cases, I arrived at a set of general
guidelines for correct redescriptions of acts in terms of
their consequences (conditions Cl-CS); these results I sum—
marized at the end of Chapter II, section 1. Finally, I
indicated the importance of these results for ethical
theory; in particular, I employed the analysis in Chapter
II to preserve the viability of the deontological-
teleological distinction in ethical theory against those
who wish to argue that such a distinction is trivial and
unilluminating.

Finally, in Chapter III I addressed myself to a
very specific question concerning the relation between act
and intention. I set out to determine (a) the relation
between an agent‘s knowledge or awareness of what he is
doing (or has done) and the concept of intentional action,
and (b) the bearing of this relation on the problem of
act-description. The results of this inquiry are summar-
ized in the concluding section of the chapter, so I need

not repeat them here.

224

The sc0pe of the problem of act-descriptions as I
have formulated it is so broad that any attempt to deal
with it in a wholly satisfactory manner within the compass
of a single volume is destined to fail. The task of apply-
ing the analysis suggested to all the sorts of cases which
must be handled is a painstaking one, and one that must be
approached in a piecemeal fashion. Each problem addressed
raises further questions; within the problem of intention—
ality, as I indicated, there arise a whole host of distinct
issues to be dealt with. We must content ourselves with
approaching each of these one at a time, exercising the
care and patience necessary for arriving at an adequate
treatment in each instance. In this respect, the sense of
frustration engendered by the incompleteness of the task
at times almost seems sufficient to outweigh whatever com—
fort or satisfaction has been attained by the progress made.
But this is as it should be. Philosophy, as has been re-
marked so often, is an ongoing activity; the sense of
frustration noted above is one of the occupational hazards
of doing philosophy. But although one must resist at all
costs the insidious temptation to rest content with a job

only partly completed, one should also resist with equal

225

force deprecating contributions--however small--to the
solution of important problems. It is in this spirit that

I, and I hope others, regard the results of this study.

BIBLIOGRAPHY

BIBLIOGRAPHY

Anscombe, G. E. M. Intention (2nd ed.). Ithaca: Cornell
University Press, 1963.

 

Atwell, John E. "The Accordion Effect Thesis," The Philo-
sophical Quarterly, Vol. 19, No. 77 (October, 1969L
337-342.

 

 

Audi, Robert. "Intending," The Journal of Philosophy,
Vol. LXX, No. 13 (July 19, 1973), 387-403.

 

 

 

Aune, Bruce. "Intention and Foresight," The Journal of
Philosophy, Vol. LXIII, No. 20 (October 27, 1966),
652-654.

Austin, John. Lectures on Jurisprudence. London: John
Murray, 1873.

 

Austin, John L. Philosophical Papers (ed. G. J. Warnock
and J. O. Urmson). Oxford: Clarendon Press, 1962.

 

Baier, Annette C. "Act and Intent," The Journal of Phil-
OSOphy, Vol. LXVII, No. 19 (October 8, 1970),
648-658.

 

Bedford, Erroll. "Intention and Law," The Journal of
Philos0phy, Vol. LXIII, No. 20 (October 27, 1966),
654-656.

 

 

Castafieda, Hector-Neri. "Intentions and the Structure of
Intending," The Journal of Philosophy, Vol. LXVIII,
No. 15 (August 5, 1971), 453-466.

 

Chisholm, Roderick. "The Structure of Intention," The
Journal of Philosophy, Vol. LXVII, No. 19 (Oc-
tober 8, 1970), 633-647.

 

226

227

Churchland, Paul M. "The Logical Character of Action-
Explanations," The Philosophical Review, Vol. LXXIX,
No. 2 (April, 1970), 214-236.

 

Cody, Arthur B. "Can a Single Action Have Many Different
Descriptions?," Inquiry, Vol. 10, No. 2 (Summer,
1967), 164-180.

. "Is Human Action a Category?," Inquiry, Vol. 14,
No. 4 (Winter, 1971), 386-419.

D‘Arcy, Eric. Human Acts: An Essay in their Moral Evalua-
tion. Oxford: Clarendon Press, 1963.

 

Davidson, Donald. "Actions, Reasons, and Causes," The
Journal of Philospphy, Vol. LX, No. 23 (November 7,
1963), 685-700.

 

Donnellan, Keith. "Knowing What I Am Doing," The Journal
of Philosophy, Vol. LX, No. 14 (July 4, 1963),
401-409.

 

 

Feinberg, Joel. "Action and Responsibility," in Alan R.
White (ed.), The Philosgphy of Action (Oxford:
Oxford University Press, 1968), 95-119.

 

Fleming, Brice Noel. "On Intention," The Philosophical
Review, Vol. LXXIII, No. 3 (July, 1964), 301-320.

 

Geach, Peter T. "Ascriptivism," The Philosophical Review,
Vol. LXIX, No. 2 (April, 1960), 221-225.

 

Gustafson, Donald. "The Range of Intentions," Inquiry,
Vol. 18, No. 1 (Spring, 1975), 83-95.

Hampshire, Stuart. Thought and Action. New York: Viking
Press, 1959.

 

Kenny, Anthony. Action, Emotion, and Will. London: Rout-
ledge & Kegan Paul, 1963.

 

. "Intention and Purpose," The Journal of Phil-
osophy, Vol. LXIII, No. 20 (October 27, 1966),
642-651.

228

Leonard, Henry S. "Essences, Attributes, and Predicates,"
Proceedings and Addresses of the American Philo-
sophical Association, Vol. XXVIII (1963-64).

Macklin, Ruth. "Actions, Consequences, and Ethical Theory,"
The Journal of Value Inquiry, Vol. I, No. 1 (Spring,
1967), 72-80.

. "A Rejoinder," The Journal of Value Inquiry,
Vol. I, No. 2 (Summer, 1967), 135-138.

Margolis, Joseph. Values and Conduct. Oxford: Clarendon
Press, 1971.

Meiland, Jack W. The Nature of Intention. London:
Methuen & Co., Ltd., 1970.

Melden, A. I. "Action," The Philosophical Review, Vol. LXV,
No. 4 (October, 1956), 523-541.

. Free Action. London: Routledge & Kegan Paul,
1961.

 

Miller, Arthur R. "Correct vs. ‘Merely True‘ Act-

Descriptions," Inquiry, Vol. 17, No. 4 (Winter,
1974), 457-460.

. Review of Action: An Anaiysis of the Concept by
David Rayfield, The Michigan Academician, Vol. VII,

Oldenquist, Andrew. "Choosing, Deciding, and Doing," in
Paul Edwards (ed.), The Encyslopedia of Philosophy
(New York: Macmillan, 1967), Vol. 2, 96-104.

 

O‘Shaughnessy, Brian. "Observation and the Will," The
Journal of Philosophy, Vol. LX, No. 14 (July 4,
1963), 367-392.

 

Pitcher, George. "‘In Intending‘ and Side Effects," The
Journal of Philosophy, Vol. LXVII, No. 19 (Oc-
tober 8, 1970), 659-668.

 

229

 

Pitcher, George. "Hart on Action and Responsibility,"
The Philosophical Review, Vol. LXIX, No. 2 (April,
1960).

Rescher, Nicholas (ed.). The Logic of Decision and Action.
Pittsburgh: University of Pittsburgh Press, 1967.

 

Rayfield, David. Action: An Anaiysis of the Concept. The
Hague: Martinus Nijhoff, 1972.

 

. "Action," Nous, Vol. II, No. 2 (May, 1968),
131-145.

. "On Describing Actions," Inquiry, Vol. 13,
Nos. 1-2 (Summer, 1970), 90-99.

"Cody‘s Categories," Inquiry, Vol. 16, No. 4
(Winter, 1973), 419-428.

. "On Miller‘s Paradoxes and Circles," Inquiry,
Vol. 17, No. 4 (Winter, 1974), 461-464.

Taylor, Charles;~ The Explanation of Behavior. London:
Routledge & Kegan Paul, 1964.

 

Taylor, Richard. Action and Purpose. Englewood Cliffs,
N.J.: Prentice-Hall, 1966.

 

White, Alan R. (ed.). The Philosophy of Action. Oxford:
Oxford University Press, 1968.

 

Wilson, J. R. S. Emotion and Object. Cambridge: Cam-
bridge University Press, 1972.

 

Yolton, John W. "Act and Circumstance," The Journal of
Philosophy, Vol. LIX, No. 13 (June 21, 1962),
337-350.

 

 

. "Ascriptions, Descriptions, and Action Sen-
tences," Ethics, Vol. 67, No. 4 (July, 1967),
307-310.

230

General References

 

Allen, Harold J. "A Logical Condition for the Redescrip-
tion of Actions in Terms of Their Consequences,"
Journal of Value Ihquiry, Vol. I, No. 2 (Summer,
1967), 132-134.

 

Anscombe, G. E. M. "Intention," Proceedings of the Aris-
totelian Society, Vol. LVIII (1956-57), 321-332.

 

 

. "Two Kinds of Error in Action," The Journal of
Philosophy, Vol. LX, No. 14 (July 4, 1963), 393-401.

 

 

Ardal, P. S. "Motives, Intentions, and Responsibility,"
The Philosophical Quarterly, Vol. 15, No. 59 (April,
1965), 146-153.

 

 

Bennett, Daniel. "Action, Reason, and Purpose," The Journal
of Philosophy, Vol. LXII, No. 4 (February 18, 1965),
85-96.

Bennett, Jonathan. "Whatever the Consequences," Analysis,
Vol. XXVI, No. 3 (January, 1966), 83-102.

Benson, J. "The Characterization of Actions and the Vir-
tuous Agent," Proceedings of the Aristotelian
§sciety, Vol. LXIII (1962-63), 251-266.

 

Brand, Myles (ed.). The Nature of Human Action. Glenview,
Illinois: Scott, Foresman and Company, 1970.

 

Brown, D. G. Action. Toronto: University of Toronto
Press, 1968.

Casey, John (ed.). Morality and Moral Reasoning. London:
Methuen, 1971.

 

Chisholm, Roderick. "The Descriptive Element in the Con-
cept of Action," The Journal of Philosophy, Vol.
LXI, No. 19 (October 5, 1964), 613-625.

 

Chopra, Y. N. "The Consequences of Human Action," Proceed-
ings of the Aristotelian Society, Vol. LXV (1964-
65), 147-166.

 

231

Cody, Arthur B. "A Reply to Mr. Dowling," Inquiry, Vol. 10,
No. 4 (Winter, 1967), 449-452.

Cohen, Michael. "The Same Action," Proceedings of the
Aristotelian Society, Vol. LXX (1969-70), 75-90.

 

 

Dowling, R. E. "Can an Action Have Many Descriptions?"
Inquiry, Vol. 10, No. 4 (Winter, 1967), 447-448.

Griffen, James. "Consequences," Proceedings of the Aris-
totelian Society, Vol. LXV (1964-65), 167-182.

 

 

Hare, R. M. "Descriptivism," Proceedings of the British
Academy, 1963, 115-134.

 

Hart, H. L. A. "The Ascription of Responsibility and
Rights," Proceedings of the Aristotelian Society,
Vol. XLIX (1948-49), 171-194.

 

Hart, H. L. A. and Hampshire, Stuart. "Decision, Inten-
tion and Certainty," Mind, Vol. LXVII, No. 265
(January, 1958), 1-12.

Heath, P. L. "Intention," Proceedings of the Aristotelian
Society Supplementary Volume (1955), 147-164.

 

 

Jager, Ronald. "Describing Acts Owing to Ignorance,"
Analysis, Vol. XXVII, No. 5 (April, 1967), 163-167.

Jenkins, J. S. "Motive and Intention," The Philosophical
Quarterly, Vol. 15, No. 59 (April, 1965), 155-164.

 

 

Ladd, John. "The Ethical Dimensions of the Concept of
Action," The Journal of Philosophy, Vol. LXII,
No. 21 (November 4, 1965), 633-645.

 

Langford, Glenn. Human Action. New York: Anchor Books,
1971.

 

Olsen, C. "Knowledge of One's Own Intentional Actions,"
The Philosophical Quarteriy, Vol. 19, No. 77
(October, 1969), 324-336.

 

232

Passmore, John. "Intentions," Proceedings of the Aristo-
telian Society Supplementary Volume (1955), 131—146.

Sikora, R. I. "Unforeseeable Consequences," Analysis,
Vol. XXIX, No. 3 (January, 1969), 85-91.

Ware, Robert. "Acts and Action," The Journal of Philosophy,
Vol. LXX, No. 13 (July 19, 1973), 403-418.

Will, F. L. "Intention, Error, and Responsibility," The
Journal of Philosophy, Vol. LXI, No. 5 (February 27,

1964), 171-179.