INFORMATION TO USERS
This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted.
The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction.
1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity.
2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame.
3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again — beginning below the first row and continuing on until complete.
4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced.
University Microfilms
300 North Zeeb Road Ann Arbor, Michigan 481 OS
A Xerox Education Company 73-1940
BEELICK, Donald James, 1941- CONTEXT AND HUMAN ACTIONS.
The Ohio State University, Ph.D., 1972 Philosophy
University Microfilms, A XEROX Company, Ann Arbor, Michigan
© 1972
•Donald James Beelick
ALL RIGHTS RESERVED
THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED. CONTEXT AND HUMAN ACTIONS
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School -of The Ohio State University
By
Donald James Beelick, B.A.
******
The Ohio State University 1972
Approved by
Adviser fj Department of Philosophy PLEASE NOTE:
Some pages may have
indistinct print.
Filmed as received.
University Microfilms, A Xerox Education Company ACKNOWLEDGMENTS
I wish to acknowledge my deeply felt indebtedness to all the people who have been instrumental in the production of this philosophical enterprise. I owe much grati tude to the members of my reading committee;
Professor Andrew G. Oldenguist, Professor
Lee Brown, and Professor William Lycan, for their excellent criticisms and comments. Any errors to be discovered in this dissertation are solely my responsibility. I am expecially indebted to my adviser, Professor Andrew G. Oldenquist, for his patient and excellent guidance in seeing this essay through to its completion. I also wish to thank Misses Cecilia Staas and Molly Bordonaro, whose fine typing work was much appreciated. Finally
I am grateful to my wife, Carol, for all the en couragement and sacrifices which made it all possible VITA
March 16, 1941 . . . Born - Kalamazoo, Michigan
1964 ...... B.A., Western Michigan University, Kalamazoo, Michigan
1963-1967 ...... Teaching Assistant, Department of Philosophy, The Ohio State University, Columbus, Ohio
1967-1972 ...... Instructor, Department of Philosophy, Wright State University, Dayton, Ohio
FIELDS OF STUDY
Major Field: Philosophy
Studies in Epistemology. Professor Richard Severens.
Studies in Ethics. Professor Andrew G. Oldenquist.
Studies in Logic. Professors Stephen F. Barker and Charles Kielkopf.
Studies in Metaphysics. Professor Everett J. Nelson.
Studies in the Philosophy of Science. Professor Virgil G. Hinshaw, Jr.
Studies in Aristotle. Professor Robert G. Turnbull.
Studies in Kant and Plato. Professor Marvin Fox.
Studies in Wittgenstein. Professor Morris Weitz. PREFACE
This essay is a discussion of the concept of human action against the background of two major questions: the question of what a human action is, and the question of whether reason explanations are a species of causal ex planations . With regard to the first of these issues it is argued that although there is no one all-inclusive definition of a human action, the notion of a basic bodily act can be adequately defined. It is also argued that human acts can be divided into two major classes: bodily acts and acts which are, in one way or another, generated from other acts. With regard to the second question, it is argued that explanations in terms of reasons are a species of scientific explanations, in that wants and beliefs cause basic human actions and hence indirectly cause all human actions.
Chapter One is an initial examination of some of the major difficulties and ramifications associated with the philosophy of action.
'fc t Chapter Two deals primarily with a number of various attempts on the part of philosophers at analyzing the concept
iv of a human action. First, a form of naive behaviorism is considered and rejected as inadequate. Next, the voli tional theory is examined. And while it is argued that many, of the stock objections fail to be conclusive, the theory is practically useless since no volitional theorist has succeeded in explaining what a volition is nor is there any independent evidence for the existence of such things as volitions. Ascripticism is then examined and it is argued both that it does not succeed in explaining what a human action is and it does not, as many of its proponents claim, rule out causal explanations of human actions. Next, Paul Ziff's version of contextualism is examined, and while it is admitted that some acts are distinguished from (mere) happenings on the basis of con text, the distinction between bodily acts and bodily movements in not entirely a contextual one, and Ziff's emphasis on the subsequent behavior of the agent is simply too vague and limiting to be of much real help in clari fying the contextual elements which are involved in some actions. Finally, A. L. Melden's and A. R. Louch's efforts at further elucidating these contextual elements is exposed. There it is claimed that the feature which seems to be central to these action theorists' attempts at elucidating the notion of a human action and a reason for acting seems to be the characteristic of consciousness, which makes the final resolution of the problem dependent on a satisfactory solution to the mind-body problem.
It is also argued that although Melden and Louch are of the opinion that our everyday concepts of persons and human actions involve certain evaluative elements which consequently rule out the possibility of giving purely scientific accounts of them, their arguments are not sufficient to demonstrate this.
In Chapter Three I deal with the supposed contrast between rational explanations and scientific explanations and the two major arguments used to show that reasons cannot be causes. The supposed contrast between rational explanations and scientific explanations is examined against the background of the debate between William Dray and Carl Hempel concerning the evaluative or empirical status of "principles of action" and it is claimed that such principles cannot be considered to be purely analytic since we sometimes abandon the hypothesis that the agent in question is rational. The appearance of analyticity attaches to Dray's "principles of action" only because of a certain epistemological interdependence of belief and goal ascription. Next, with regard to the two major arguments which have been presented to show that reasons canno.t be causes: the so-called "logical connection argument, and the argument that reasons cannot be causes because they are known in a radically different way than causes, I argue first that the supposed logical connection
between reasons and actions is more grammatical than logi
cal. Philosophers who have supposed that there is a
logical connection between reasons and actions have merely
confused the intentional object of a reason with some
actual objects or acts. Concerning the second argument,
it is argued that the notion of knowledge without ob
servation is simply too vague and obtuse to establish
that reasons are not causes, and even if reasons are
known in a different way than causes, this is not suf
ficient to show that reasons are not, in reality, causes.
Finally, it is claimed that some philosophers have failed
to see that the defining characteristic of all actions is not simply that they are bodily movements in certain cir
cumstances, because they have failed to see that there
are at least two distinct senses in which actions are contrasted with bodily movements.
In Chapter Four, I argue that some philosophers have been committed to a number of incompatible theses con cerning human actions and how they may be correctly de scribed. The incompatible theses in question are these:
(1) a set of act descriptions, e.g., "John flipped the switch," "John turned on the light," "John illuminated the room," etc., all describe one identical action; (2) a reason may succeed in explaining an action under one
vii description but fail to explain it under some other description; (3) reason explanations are a species of causal explanations. Given this inconsistency X argue
further that the first thesis is false and must be re jected. The rejection of this thesis consequently requires the formulation of more precise techniques for individuating actions, and the time-honored type-token distinction is suggested as a viable way of doing this.
And finally, it is argued that this type-token distinction has the advantage of allowing us to explicate more fully the structural relationships between various acts con tained in our everyday use of the "by" and "in" locutions.
Next, in Chapter Five, G. E. M. Anscombels theory of intentional action is presented and rejected on the grounds that it depends upon her obscure notion of knowledge with out observation. In view of the failure of this account of intentional action, a preliminary justification of a want-and-belief causal account of intentional action is presented. Next, Alvin Goldman's theory of intentional action is presented and, except for a minor revision in his definition of a basic act-token, defended. And finally,
Goldman's hypothetical account of human ability is defended against the objections of J. L. Austin.
In Chapter Six, the nature of wants and beliefs is explored. First, the distinction between occurrent and
viii dispositional wants and beliefs is drawn. It is also pointed out that there is nothing in the logical analysis of belief and want sentences which would prevent beliefs and wants from being considered as causes. Goldman's argu ments for the supposed mental status of wants and beliefs is then examined and rejected. After this, explanations in terms of wants and beliefs is compared with explanations given in the behavioral sciences, and it is argued that explanations in terms of molar behavior leave much to be desired. Next, the question of the incompatibility of neurophysiological explanations and want-and-belief explanations of human actions is explored and I argue that, because the possibility of a plurality of causes cannot be discounted, the discovery of neurophysiological causes for human behavior does not invalidate explanations of them in terms of mental causes. And finally, it is argued that in spite of this, there are no insurmountable barriers to the possibility that wants and beliefs could be identified with neurological events, states, and/or structures even though belief and want sentences are not synonymous with the sentences describing these neurological events, states, and structures. TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS ...... ii
VITA AND FIELDS OF S T U D Y ...... iii
PREFACE ...... iv
Chapter
I. INTRODUCTION: THE NATURE OF THE GENERAL PROBLEM ...... 1
II. THE NATURE OF...... HUMAN ACTION ...... 10
2.1. What Needs to be Explained .... 13 2.2. The Supposed Common Feature of A c t i o n s ...... 17 2.3. The Supposed Unanalyzability of The Concept of a Human Action . . 20 2.4. Mental Acts ...... 27 2.5. Naive Behaviorism ...... 29 2.6. The Volitional Theory: A Human Action is a Bodily Movement Caused by a Volition ...... 33 2.7. Ascripticism ...... 50 2.8. Contextualism...... 78 2.9. A.I. Melden's T h e o r y ...... 82
III. EXPLANATIONS, REASONS, AND CAUSES .... 116
3.1. Rational Explanation and Scien tific Explanation...... 118 3.2. Belief and Reason Explanation . . . 125 3.3. Dray's Principles of Action .... 130 3.4. Reason Concepts as Dispositions . . 140 3.5. The "Logical Connection" Argument . 143 3.6. Knowing Our Reasons Versus Knowing C a u s e s ...... 158 3.7. Two Distinct Approaches to the Problem of Human Actions .... 170
x IV. ACT DESCRIPTIONS AND ACT GENERATION . . . 173
4.1. Action Descriptions ...... 176 4.2. "Doing X by Doing Y" And "Doing X in Doing Y " ...... 185 4.3. Basic Actions Vs. Nonbasic A c t i o n s ...... 192 4.4. Action Types and Action Tokens . . 195 4.5. Act G e n e r a t i o n ...... 203
V. A CAUSAL ACCOUNT OF HUMAN ACTION .... 214
5.1. Intentional A c t s ...... 215 5.2. The Justification of a Causal Account of Intentional Action . . 226 5.3. Goldman's Account of Intentional A c t i o n ...... 230 5.4. Hypothetical Accounts of Human A b i l i t y ...... 247
VI. WANT AND BELIEF - EXPLANATIONS AND MECHANISM...... 263
6.1. Wants and B e l i e f s ...... 264 6.2. The Supposed Mental Status of Wants and Beliefs ...... 279 6.3. Behaviorism...... 282 6.4. Neurophysiological Causes as Rival Explanations of Human Actions . . 290
VII. CONCLUSIONS...... 308
BIBLIOGRAPHY ...... 314
xi CHAPTER I
INTRODUCTION: THE NATURE OF THE GENERAL PROBLEM
It would appear that there are fewer notions that we are more at home with in our everyday discourse about our
selves and others than that of a human action, or what a person does. In everyday parlance it seems clear that men act and bodies behave, and, what may not be the same thing, that an important distinction is to be made between what a man does and what happens to him. Indeed, this distinction is so common that it is quite natural for us to take it for granted as presenting no great mystery. But although many less reflective persons have been perfectly satisfied to leave this distinction at the intuitive level, philosophers have been sufficiently struck by its seeming importance and mystery to attempt to discover its precise nature. For better or worse, I have found myself sufficiently perplexed over this distinction so as to be troubled by it, and hence the following represents my attempt to discover whether and to what extent it is possible to find adequate criteria for drawing this distinction.
Perhaps it would be advantageous at this stage to raise the question of the philosophical importance of such an
1 investigation. That is, it may well be asked exactly what philosophical problems can one reasonably expect such an investigation to solve, or if it will not in itself solve any philosophical problems, then what particular philosophical problems can such an investigation be expected to give us additional insights into. First, I do not think that anyone who has taken an active and reflective part in ethical disputes can deny the philosophical significance of attempting to make this distinction clearly and precisely. I, for one, have certainly felt the lack of any such criteria. Debates in ethics commonly and naturally give rise to the question of what is to be considered as the act of an agent as opposed to (mere) happenings, the circumstances of his act, and the consequences of his act. Assuming that when we are concerned with questions of rightness, wrongness, obligatoriness, and permissibility we are primarily concerned with judging an agent's actions, the task of providing criteria by means of which actions may be distinguished from (mere) happenings, circumstances, and consequences becomes of cardinal importance.
For, if we are never quite sure about exactly what is to be construed as an agent's act as opposed to these other things, then we will find ourselves in the embarrassing position of talking past each other. It would seem then that no normative theorist can hope to establish that his theory is somehow better than his opponent's unless there is some kind of initial agreement about exactly what it is that is being talked about. Another philosophical problem that should stand to gain some clarity from an investigation into the nature of human actions is the problem of determinism. Although it is debatable whether common sense uses the distinction between what a person does and what happens to him as a bulkhead against determinism, no one who is familiar with the literature can deny that this seems to be one of the purposes many philosophers have had in mind in insisting on the distinction. Indeed, some philosophers have been so convinced of a radical difference between human action and
(mere) bodily behavior that they have either denied that there can in principle be a "science" of psychology or sociology patterned after that of the physical sciences such as physics and chemistry, or at least have raised serious questions about the possibility of adopting the methods of the physical sciences in these disciplines.
The exact nature of my problem is to determine to what extent it is possible to discover clear and precise criteria for distinguishing human actions and mere physical movement, and also to what extent are the claims that human actions are uncaused justified. The first part of my problem may be interpreted in at least three different senses. First, it may be interpreted as a search for definition of "action," i.e., a search for the necessary and sufficient conditions for the correct application of the term. Secondly, it may be conceived as a search for a contextual definition, i.e., a search for a statement or set of statements which is logically equivalent to the statement, "S performed the act A." Fail ing either of these, my problem may be interpreted as a search for an elucidation of the use of the word "action."
The distinction between human actions and mere physical movements most likely has its roots in ordinary discourse, even though perhaps unclearly. But this fact cannot be in terpreted as legislating against departing from ordinary usage. In fact, we must depart from it whenever it is muddled and likely to mislead us, and it may very well be that many of the expressions which are used to make the distinction between human actions and (mere) happenings or bodily movements are vague and liable to mislead. If this should turn out to be the case, then there can be no objection to replacing them with new technical terms for our own use. No doubt, the best theory of human action would be one which had the characteristics of simplicity and completeness, perhaps ruling out a few cases as
"abnormal." But we must not presuppose that there is a unity in the variety of uses that the term "action" or any of its equivalents have in ordinary discourse, and hence, we must not presuppose that any general theory of human action as such is possible.
In fact, it will be a major part of my thesis that there is simply no all-encompassing analysis of our everyday concept of a human action. That is, I shall argue that it is impos sible to specify exactly what the necessary and sufficient conditions for the correct application of the term "human action" and its equivalents are, nor do I believe that any completely satisfactory contextual definition of that term can be given. This thesis, however, needs to be qualified to some extent. For, as I shall argue, the notion of a basic bodily action can be defined, i.e., the necessary■and sufr1 ficient conditions for the correct application of the term
"a basic bodily action" can be given. The question of whether or not this should be called an analysis of the concept of a basic bodily action is complicated by the fact that at least some of these analysans themselves refer to other concepts for which I am unable to provide exact criteria for their correct application. This situation is not unlike the case in which the concept of a voluntary human action is first
"analyzed" in terms of an intentional human action which is in turn "analyzed" in terms of some further unanalysable concept like "circumstances." Should we, in this case, say that the concept of a voluntary human action has been successfully analyzed or not? In this case it seems to me that the issue of whether or not the concept of a vol untary human action has been analyzed or not is merely a verbal one, and therefore I will continue to claim that whereas the general concept of a human action cannot be successfully analyzed, the concept of a basic bodily action can be.
It would appear that the most philosophically trouble some problem associated with the philosophy of human action, and the one occasioning the most lively debate has been the problem of what constitutes a correct explanation of human action. In recent philosophy, a number of prominent philo sophers (A.R. Louch, A.I. Melden, R.S. Peters, R. Taylor, et al.) have explored the logical differences between how the concepts which are used in our everyday discourse about human actions function and the way in which the concepts of the natural sciences function. One of their major theses has been that these concepts (e.g., motives, intentions, purposes, acts, and actions, etc.) behave in radically dif ferent ways from the way in which the concepts of the natural sciences behave; they are not, for instance, concepts which can be explicated in terms of events and processes. One
extension of this thesis is the claim that human actions
are just not the sorts of things of which it makes sense
either to ascribe or deny a cause to. On the other hand, other philosophers (D. Davidson, J. Podor, C. Hempel,
et al.) have argued that this thesis is basically mis
taken and that either these concepts are relics of a prescientific era or that they can be fully explicated
in terms of laws. The debate, therefore, most frequently has taken the form of providing an answer to the question,
"How are human actions to be explained?" with some phil osophers arguing that a complete explanation of human actions in terms of laws relating events is possible while others deny it.
The champions of the covering-law theory of explan ation have insisted that all true explanation, all attempts at explaning a phenomenon which deserves the name of explanation, must be ones which proceed in terms of the deduction of that phenomenon from laws or law-like state ments. Thus, they conceive of the proper task of the psychologist and sociologist as the discovery of laws or law-like statements which would provide such explanations
* of human actions. An example of such a law-like statement of sociology might be, "The property of expressing group loss which characterizes ritual wailing in primitive societies is sufficient to produce group cohesion and solidarity which is, in turn, a necessary condition for the maintenance of survival of the society." But the truth of the matter is that scarce few such "psychological laws" or "laws of social development" have been discovered so far, and it does not appear that there are likely to be many more in the immediate future. Generally, the defenders of the covering-law theory have rebuked this claim, claiming that there really is no reason why such laws cannot be successfully formulated and used. What is needed most of all, they claim, is a Newton for the social sciences who will finally set things straight. In other words, the defenders of the covering-law theory claim that the relatively limited success psychologists and sociologists have had in discovering such laws in the past is due to the fact that the variables involved are simply too complex for any simple-minded generalizations.
But the difficulty here, the critics of the covering- law model of explanation have argued, is not that no one with sufficient genius has occurred on the scene to finally set things straight, but rather that the program prescribed by the covering-law theory is too limiting. The writings of certain philosophers of science like Ayer, Hempel,
Bergmann, et al. have been interpreted by some psycholo gists and sociologists as providing the methodological criteria for any science of psychology and sociology,^" and thus, many psychologists and sociologists have con sidered it their aim to pattern their inquiries after the manner of physics, and see as the object of their investigations observable and measurable behavior which can be formulated into laws which state relationships between two or more variables.
The difficulty with this methodology is not so much one of prospect as it is one of practice. For, as outlined above, the philosopher can have little objection to this program. The history of philosophy is laden with the
1 The philosophers who support this position are quite numerous. Cf. e.g. R.B. Braithwaite, Scientific Explanation (Cambridge: Cambridge University Press, 1964); John Hospers, "What Is Explanation?" in Essays in Concep tual Analysis, ed. by Antony Flew (New York: St. Martin's Press Inc., 1956), pp. 94-119; Carl G. Hempel, Philosophy of Natural Science (Englewood Cliffs, New Jersey: Prentice- Hall, Inc., 1966), pp. 47-110; Ernest Nagel, The Structure of Science (New York: Harcourt, Brace and World, Inc. , 19*61); Richard S. Rudner, Philosophy of Social Science (Englewood Cliffs, New Jersey: Prentlce-Hall, Inc., 196*67"; Carl Hempel and Paul Oppenheim, "The Logic of Explanation," in Readings in the Philosophy of Science, ed. by H. Feigl and M. Brodbeck (New York: Appleton-Century-Crofts, Inc., 1953), pp. 319- 352. 10
skeletons of philosophers' attempts to tell the scientist what he can and cannot do. But although there is no
objection to such a proposal in principle, there can and has been an objection to the way in which the proposal has been put into practice. For in practice, many psycholo gists and sociologists claim that what they are explaining and predicting is not merely the various bodily movements that human beings make but rather, the actions they per form. But the question immediately arises of whether human actions can be reduced in the required way to ob servable and measurable behavior. The great power of natural science lies in its ability to predict, and hence, its ability to manipulate phenomena. The only question is whether the psychologist and sociologist can adopt the methodology of the natural sciences wholesale and still claim, as some of them apparently do, that they are giving causal or nomological accounts of human actions.
Some philosophers have thought that they have seen in the arena of human action a citadel in which human freedom might defend itself against the threat of a com plete determinism. My view, however, is that such hopes are not justified. Valuable lessons are to be learned about how these various action concepts function in our language from what these philosophers have to say. But, 11
I fear that the lessons to be learned are not sufficient to justify the claim that determinism has thereby been refuted. And, as we shall come to see, the lessons to be learned from these philosophers are of quite a dif ferent order.
In the course of what follows I shall argue that there are at most two plausible senses in which we may speak of human actions. First, the sense in which a bodily action is opposed to a (mere) bodily movement is the sense in which actions are bodily movements which are caused by the agent's wants and beliefs. The second sense in which we may plausibly speak of a human action, the sense in which actions are being contrasted with happenings other than (mere) bodily movements, is the o sense in which actions are "generated" either from bodily actions or else are generated from generated acts.
I shall also argue that all human actions are caused, and that a completely adequate hypothetical account of a person's ability to perform an action can be given.
Finally, I shall defend the thesis that there is no
2This technical notion will be explained thoroughly in Chapter IV. ; insurmountable impasse between our common-sensical conceptual scheme and the scientific scheme, nor is there any plausible reason why our common-sensical conceptual scheme cannot be successfully superseded by the scientific scheme. CHAPTER II
THE NATURE OF HUMAN ACTION
2.1. What Needs to be Explained
Any philosophical theory which seeks to be something more than an exercise in intellectual wool-gathering aims at providing some sort of a coherent account of some set of facts or phenomena. The question which naturally suggests itself with regard to the philosophical theory of human action then is "exactly which facts or phenomena is any theory of human action obligated to explain?" The answer to this question would seem to be in one way quite obvious, but I think that it can be shown to be a great deal more difficult and perplexing than originally thought.
The obvious answer is, of course, simply actions, and in particular, human actions. Yet once it is granted that we are searching for an answer to the question, "What is a human action?" it does not follow that it is the sort of question for which there exists any clear-cut and unambig uous answer. This is so mainly because, as I shall argue, there is no one common intuitive notion of what consitutes a human action, nor does there appear to be just one contrast
13 14 involved in many action philosophers' attempts to dis tinguish between what a person does and what merely happens to him.
In any case, once it is granted that we are looking for a precise answer to the question, "What is a human action?" we may still ask exactly what it is about human actions which needs to be accounted for. With regard to this question I do not believe that it can seriously be denied that human beings, and some animals as well, normally are able to move various parts of their bodies, and perform certain intellectual and semi-intellectual tasks as well. Normal human beings, for example, perceive things, solve certain problems, understand written and spoken language, remember certain episodes, recognize certain things, move certain bodily parts, formulate and execute plans of strategy, and many, many others. And human beings also seem to be able to perform these things in a variety of different ways. They may do them inten tionally, deliberately, for some purpose, voluntarily, by inclination, by habit, by choice, after deliberation, and so on. The facts, then, that any adequate theory of human action will have to explain are exactly these. That 15 is, any adequate theory of human action will have to explain exactly what it is for a person to perform an action, what it is for a person to be able to perform an action, and finally, what it is for a person to do something inten tionally, voluntarily, by choice, deliberately, on purpose, etc.
It may be objected that my way of putting this in volves a confusion between the task of philosophy and the task of psychology. That is, it may be objected that my way of putting things suggests that the task of a philo sophical theory of human action is to explain actions, whereas, in fact, this is the task of psychology. This objection is well-taken. It surely is not the job of the philosopher but the job of the psychologist to explain why people perform the particular actions that they do. The job of the philosopher is rather to analyze or elucidate certain facts about actions or why we talk about actions in certain ways. The difficulty here seems to lie in the word "explanation," for some action theorists have claimed that it is questionable whether and to what extent psy chologists explain human actions. Hence, Melden writes: 16
There is nothing intrinsically objectionable in radical alterations of the uses of terms bor rowed from everyday discourse; the history of the sciences affords us many examples of this phenomenon. The trouble in the case of the psychologist's prac tice is that all too frequently the radical shifts in the use of such terms as "skill,1' "learning," "action," "motive," and so on, go unnoticed . . . In one breath we are sometimes told by the psy chologist both that the everyday uses of these key terms are vague and obscure and that he is concerned to explain familiar psychological phenomena. This is a paradigm of logical incoherence. If the terms are obscure and vague, then what is equally obscure and vague is the scope of the subject-matter for which explanations are required. . . What such accounts purport to explain, is the fact that beings endowed with such developed bodily mechanisms are capable not only of moving their limbs at will but also of performing the very many sorts of things to which we have applied the term "human action": the unlocking and opening of doors, signalling, etc., etc. Here the crucial term is "explanation,11 the meaning of which must remain obscure unless and until we have become clear about the gap, . . ., between matters of physiological or bodily hap penings and matters of human a c t i o n . 3
Whether there is a viable distinction between matters of physiological or bodily happenings and matters of human
actions is, or course, debatable. But it will hardly do if we beg the issue from the beginning by assuming that psy chologists and physiologists explain human actions.
In light of the supposed difference between the task of the psychologist and the philosopher, it may be helpful
q A. I. Melden, Free Action (London: Routledge and Kegan Paul, 1965), pp. 71-72. 17
to recast the philosopher of action task as "explaining"
exactly what it means to say that a person performed an
action, what it means to say that a person is able to
perform an action, and finally, what it means to say that
a person did something intentionally, voluntarily, by
choice, deliberately, on purpose, and so on.
2.2. The Supposed Common Feature of Actions
Frequently when we come to consider some philosophical question such as "What is a human action?", and a great number of other questions as well, we naturally assume that all of us are talking about the same thing. Thus, when we debate such questions as "What is truth?", "What is meaning?", "What is mind?", "What is good?", What is reference?", "Which is (are) the best political system(s)?", and so on, we are frequently inclined to assume that there is just one thing which these things are, and it is merely a question of which one of our various answers best captures its nature. Similarly, when we broach the question,
"What is a human action?", there exists an almost over whelming tendency to assume that in some miraculous way there is only one thing which is in all circumstances a human action. To say that a person has an intuitive notion 18
of something, however, only means that he is able to give examples of it. But it does not follow from this
that his examples cannot be challenged nor that the
examples themselves have any one thing in common.
It might be suggested that our common-sensical notion of a human action suggests that an essential feature of human action is the exercise of control. That is, it would appear that the agent's exercising of his ability to control what takes place is the deciding factor in classifying what occurs as either a human action or a mere happening. If the agent controls what occurs, then he is performing the said action; and if, on the other hand, he is powerless to control what occurs, or fails to control what occurs, then it is classified as something that merely happens to him.
If, for instance, my arm moves sporadically as the result of some nervous disorder, then my arm movement is classified as a mere happening, but if my arm moves as a result of whatever normally occurs when a normal individual moves his arm, then the arm movement is classifiable as an action that
I perform.
But, in spite of the initial plausibility of such a claim, it is doubtful whether it can, by itself, bear the entire weight of the distinction between human actions and human sufferings. As I shall argue in Chapter V, the feature 19
of control or the lack of it is itself explicable in terms
of belief and want causation and that such causation is
sufficient ,to explain the relevant difference between what 4 are termed basic bodily actions and mere bodily movements.
But it is doubtful that this feature can be considered to be
a surefire way of distinguishing all actions from mere hap
penings for the simple reason that not all actions are basic
actions. Actions such as checkmating your opponent and
signalling for a left-hand turn are actions which depend on
the existence of certain conventions which the agent has no
control over, or, at least, only a very minimal control.
Furthermore, although many of the things which result as a
causal consequence of a bodily action are classifiable as
human actions, it is by no means clear to what extent we
can say that what took place was within the agent's control.
Hence, although Smith’s death which results from a richo-
cheting bullet fired by Jones may be classified as Jones'
act of killing Smith, it is far from clear that Jones can be correctly asserted to necessarily have exercised any control over the event of Smith's death. The exercise of control, then, although it is an essential feature of some
4It will be noted that this way of putting things ig nores the possibility that not all basic actions are bodily actions. The reason for this will become apparent in Section 2.4 where I discuss the possibility of mental acts. 20
human actions, can hardly be supposed to be a defining
feature of all human actions.
2.3. The Supposed Unanalyzability of the Concept Of a Human Action
Philosophers have not always been careful enough in distinguishing carefully between the concept of an action and the concept of a human action. Sometimes a philoso pher tends to speak of such properties as consciousness, intention, motive, and purpose as if they were distin guishing features of human actions.5 Yet, although we commonly speak of the "action" of the sun in bleaching the carpet and the "action" of a piston in moving up and down within a cylinder, it is doubtful whether such uses of the word "action" are meant to refer to anything more than mere movement. The notion of an action, as opposed to a human action, is certainly one of the most difficult concepts to explicate. In Aristotle's system of categories it figures as one of the eight, sometimes ten, ultimate un- analyzable notions. Given the standard interpretation this would mean that the final answer to the question, "What is it?" about a certain class of things is simply, "an action."
5K. Baier, "Action and Agent," Monist, XLIX, (1964) pp. 183-195. 21
Likewise, at least one contemporary philosopher, Richard
Taylor, has argued that the concept of a human action is g an unanalyzable ultimate notion. But if the notion of a human action is an unanalyzable ultimate notion, then nothing more illuminating can be said about it than, "There simply are human actions," or perhaps, "People perform actions." We may recognize that any consistent conceptual scheme has its primitive notions (including our natural language, if it is consistent), without at the same time agreeing that human actions are primitives in any consis tent conceptual scheme powerful enough to account for all human action.
Taylor's argument for the unanalyzability of the concept of a human action is tripart. According to Taylor, there are only three ways in which we could present an analysis of a simple statement of an act. First, we may produce a statement or set of statements which is logically equivalent to the simple act statement, i.e., one which is both logically necessary and sufficient for the simple act statement. Lacking this, we may instead produce a state ment or set of statements which is materially equivalent to the simple act statement. Or finally, we may produce
6Richard Taylor, Action and Purpose (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966). a statement or set of statements which Taylor calls
"nomically" equivalent to the simple act statement.
A statement or set of statements is nomically equiva
lent to another statement provided that it would warrent
counterfactual inferences either way. Although Taylor
says either way, judging from what he says, he means
both ways. That is, if the simple act statement were
false, then the nomic equivalent(s) would be false also;
and if the nomic equivalent(s) were false, the simple
act statement would be false also. Nomic equivalents are
necessary and sufficient for the truth of the simple action
statement but still logically independent of it.
Now consider the simple act statement "X move my hand"
call it (1). A material equivalent to (1) would be, as
Taylor claims, useless as an analysis of anything since any
pair of false statements and any pair of true statements with totally unrelated subject matters are materially equi valent, Likewise, a logical equivalent of (1) would be equally of no philosophical interest since if, for instance
(1) is analyzed as either "my hand moved, and this motion was my act," or "my hand moved and I caused this motion," and so on, it is compatible with there being no such thing 23 which causes the motion of the hand. But such a logical
analysis of (1) is not a legitimate insight into the
nature of human action since it only tells us that a certain pair of expressions has the same meaning in some language or
other. Taylor, however, has not shown by this argument that
no logical analysis of (1) is possible, nor has he shown that a logical analysis of (1) would be unenlightening. For, although his own proposed logical analysis of (1) is not a legitimate insight into the nature of human action as such, it is nevertheless enlightening to discover that a certain pair of expressions share the same meaning.
A proposed nomic equivalent of (1) would be "my hand moved, and this motion was caused by an event of kind x."
Taylor argues that although this is the only kind of analy sis which would be philosophically interesting in that it would shed some light on the concept of a human action, such an analysis cannot possibly be given. Taylor has two argu ments for this which, I believe, are really just two versions of the same argument. His first argument is that it is perfectly possible that the analysands is true while (1) is false.
It will always be perfectly possible, in other words, that my hand moves as a result of the occurrence of that physical event x within me, 24
which we can suppose to have been brought about by an electrical impulse, drug, or what not, but that I did not move it, or have anything to do with its moving./
The second reason why no such nomical analysis is possible,
Taylor explains, is
. . . the event for which we allow x to stand, whether this be mental of physical, will always have to be understood as something that I do if the analysis is to have any plausibility at all. . . So long as we can imagine that some such event occurred, perhaps as a result of being artificially induced in me by electrical impulses or drugs, and that it thus occurred in spite of me and that I had nothing to do with bringing it about, there is hardly any temptation to regard its occurrence, together with the oc currence of the bodily motion in question, as sufficient for the truth of (1).
Both of these arguments, however, assume that whatever may
cause me to move my hand it must be something that is in
the strict sense something I do in order for my hand move ment to be an act of mine. And in this sense, they resemble
Melden's arguments to the effect that if volitions are to
succeed in explaining the difference between a mere bodily movement and a bodily action, they must be something that
^Ibid., p. 94.
^Ibid., p. 95. 25
I do rather than things which just occur within me.^
But this assumption is unwarranted. The primary reason is that there is no incompatibility between event- causation and object-causation. That is, to say that an agent caused something to happen is not only not incompatible with saying that some state or event in volving that agent caused that something to happen, it implies it. And secondly, there is no reason why the event v/hich eventually causes my arm to move must be something that I do in order for my arm movement to be my act. Taylor has failed to show that the relevant dis tinction between a mere bodily movement and a bodily action does not lie in a difference in the kind of event- causation involved. Bodily actions may indeed be caused by a particular class of events involving particular structures or areas of the organism while mere bodily move ments like muscle spasms and reflex movements are caused by a completely distinct set of events involving distinct structures or areas of the organism. But what causes these particular sets of events themselves is, strictly speaking, irrelevant so far as the distinction between
9A. I. Melden in "Willing," in The Philosophy of Action, ed. by Alan R. White (London, England: The Oxford Univer sity Press, 1968), pp. 71-72. 26 a mere bodily movement and a bodily action is concerned.
And even if this event or set of events is induced artificially, there is no reason to suppose that the
* subject himself will, in fact, be able to detect any phenomenological difference between this case and the cases in which these events occur naturally. That is, the bodily movements which are thus artificially induced in this fashion may appear to the subject to be just as much something that he does by his own will as when he performs this movement naturally. Of course, if the artificial stimulus in interjected at some point in the causal chain which leads from x to the arm moving, e.g., applying an electrical impulse to the efferent nerve in the arm, then the subject may be able to detect some ex periential difference between this case and the normal case. He may be amazed, for example, to see his arm moving about all by itself, so to speak. But, such an experien tial difference may not be detected if the artificial stimulus is applied at the proper place. Hence, Taylor's arguments fail to show that the concept of a human action cannot be analyzed nomically, and we will see in Chapter
IV how such an analysis might proceed. 27
2.4. Mental Acts
It may appear as though many action theorists and
their critics alike have fallen into the assumption that
if anything is going to count as a human action, it would
have to be something which is in principle open to public
observation. Evidence for this claim can be found almost
everywhere Thus, Melden has claimed:
When I perform an action, there is some bodily movement that occurs, but not every bodily move ment counts as an action - not even those of normal adult human beings - since there are reflex movements, the activities of those who walk in their sleep, and the behavior of those under hypnosis.
^This tendency, although not always explicitly stated, may be thought to be implicit in the selection of examples. See G.E.M. Anscombe, Intention, (Ithaca, New York: Cornell Univ. Press, 1969); Eric D'Arcy, Human Acts: An Essay in Their Moral Evaluation, (Oxford Univ. Press, 1963); Donald Davidson, "Actions, Reasons, and Causes," in The Nature of Human Action, ed. by Myles Brand (Glenview, 111.: Scott, Foresman and Co., 1970); H.L.A. Hart, "The Ascription of Responsibility and Rights," in Logic and Language, ed. by Antony Flew (First Series; Oxford: Basil Blackwell, 1955); May Brodbeck, "Meaning and Action," in Philosophy of Science, XXX (Oct., 1963); W.D. Gean, "Reasons and Causes," m Review of Metaphysics, XIX (June, 1966); John Benson, "The Characterization of Actions and the Vir tuous Agent," in Proceedings of the Aristotelian Society (New Series), LXIII (1962-1963); and Arthur Danto, "Basic Action," in The Nature of Human Action, op. cit. 11 A. I. Melden, "Action," Readings in the Theory of Action, ed. by Norman S. Care and Charles Landesman, (Bloomington, Ind.: Ind. Univ. Press, 1968) pp. 27-28. 28
And elsewhere,
. . . paying a grocer is not less an action, something one can observe taking place, ... . 12
However, to interpret this as meaning that Melden,
Hart, and Anscombe deny that there are any private mental acts would be a serious mistake. For, first of all, none of these theorists, in spite of their own personal views on the existence of mental occurrences, deny that mental events and acts occur. Secondly, if an action is simply the bringing about of something, there is no requirement that what is thereby brought about must be something overt.
Frankly, I do not know that there is any way of demonstrating there there either is or is not any special class of private
"mental" events or acts. If there are such acts and events, any completely adequate theory of human action would have to take them into account. But, if there is any viable distinction between what a person does and what merely happens to him, one legitimate aspect of that distinction is the supposed distinction between a (mere) bodily move ment and a bodily action, i.e., between my raising my arm and my arm rising. And it would appear that if anything could qualify as a paradigm case of a human action it would be an instance of my raising my arm.
12 A.I. Melden, Free Action, op. cit., p. 178. The major problem then is what, if anything, accounts
for the distinction between a (mere) bodily movement and a
human action. A suggestion which may not be without merit
is that we turn to what other philosophers have suggested
as ways in which this distinction may be drawn. I propose
then, to approach the problem of clarifying this distinction
between bodily movements and human actions dialectically.
That is, my procedure will be to examine the views of var
ious philosophers on the nature of human actions, presenting
refutations of what I consider to be false in their theories,
and attempting to retain any illuminating truths which might be present in them.
2.5. Naive Behaviorism
An initially plausible attempt at analyzing the con cept of human action is simply to identify the action with a piece of behavior or the behavior plus its effects. Thus, according to this view, which we might label naive be haviorism, human actions are simply bodily events and the effects of bodily events. Accordingly, a person is said to perform an action A if and only if either there occurs an appropriate piece of bodily behavior or there occurs an appropriate effect of this piece of behavior. Thus, a person raises his arm if and only if his arm rises. 30 However, this attempt at simply identifying the action with the appropriate piece of behavior is subject to numerous counterexamples. For reflex movements, nervous spasms, sleep walking, having limbs moved by other people and machines, and possibly the things done under hyp nosis are all instances in which the appropriate piece of behavior or the appropriate effect of the behavior occurs, but no corresponding act is performed by the subject.
In addition to these apparent counterexamples, A. I.
Melden has argued that any attempt to reduce human actions in the required way to a piece of appropriate behavior or the appropriate effect of this behavior is fundamentally mistaken. According to Melden, the naive behaviorist who attempts to bring off such a reduction of human actions to bodily movements is led to argue that what a person does in raising his arm is to cause certain muscles to move which, in turn, cause his arm to rise. But such a view,
Melden claims, cannot be correct because if it were true, then the agent must know which muscles to move (which mus cle movement(s) to produce) in order to produce as a result the rising of the arm, the bodily movement. In other words, he must be able to give a true account of that muscle move ment. Yet it is obvious that the agent need know nothing 31 at all about these physiological happenings in order to raise his arm. Indeed, -if this view were correct, Melden argues, only physiologists or people who had some know ledge of physiology could ever succeed in raising their arms, and this is patently false. Therefore, Melden concludes that the view that one raises his arm by moving certain muscles is mistaken, and the correct ex planation is that ordinarily one moves his arm muscles by raising his arm.-^ But in saying this Melden does not mean to claim that the arm’s rising is not caused, and is not caused by those muscle movements, but rather, that such muscle contractions do not cause the action of raising the arm.
Melden1s argument that human actions cannot be re duced to bodily movements is fallacious, however, : For it is simply false that if what a person actually does in performing the bodily action of raising his arm is to cause certain muscles to move which, in turn, cause his arm to rise, then he must know (whether this involves being able to give an account or not) which muscles to move in order to cause his arm to rise. Indeed, it is difficult to
13A. I. Melden, Free Action, op. cit., pp. 59-61. 32
imagine even the most “naive" behaviorist requiring that
people who perform such, bodily actions must have such a knowledge of physiology. The question-begging ploy of
claiming that such muscle movements cause the arm1s rising but not the action of raising the arm is frequently used by Melden. So much so, that whenever it looks as though some analysis of human actions is forthcoming, Melden almost invariably replies that what has been analyzed are
(mere) bodily movements, not actions.
Numerous examples of this ploy can be discovered quite readily in Melden*s words. Hence, in Free Action, he writes,
Grant then that wanting or desiring explains the bodily movements that take place when a person does anything, e.g. raises his arm in order to signal; as internal occurrence what it explains, at best, is the bodily movement that occurs when the person raises his arm, not the action he per forms which we describe as "raising his arm" or, further, as "signalling.
And elsewhere,
If the causal sequence is motive-bodily movement, no action is being explained at all. In order to provide the envisaged causal explanation of the action, the bodily movement has to be further identified as the bodily movement that occurs when the action is performed. Hence, in order to explain the action, for example, of my raising my arm, by
*^Ibid., p. 109. 33
citing the motive as a Humean cause of the bodily movement that consists in the rising of my arm, it is necessary to conjoin the statement of the causal relation (motive- bodily movement) with a further statement connecting the bodily movement with the action to be explained.3*5
Clearly Melden is merely assuming that there is some
distinction between a basic bodily action such as raising my arm and a mere bodily movement like my arm rising, and
that no internal cause can succeed in explaining this difference since such internal causes would only explain the arm's rising, not the action of raising the arm. Of course, Melden thinks that he has a good reason for denying that an internal cause could ever account for the difference between a mere bodily movement and a bodily action, namely, the so-called "logical connection argument." But as we will see in Chapter III, this argument is not sufficient to establish that bodily acts cannot be distinguished from mere bodily movements by a difference in internal causes.
2.6. The Volitional Theory: A Human Action is a Bodily Movement Caused by a Volition
Another view concerning the correct analysis of human actions is that human actions are bodily movements plus some
■^Ibid., pp. 84-85. C f . pp. 21, 90, and 200. 34 sort of antecedent event, usually an antecedent which is
* claimed in some sense to be internal to the agent himself.
Hence, this view suggests that although a human action is not identical with a bodily movement, it is really a bodily movement plus some internal event. Various forms of this view are of course possible depending upon how the theorist conceives of the nature of this antecedent, whether it is mental or physical, for example, and how he conceives of the relationship of this antecedent to the relevant bodily behavior. Sometimes it is alleged that this ante cedent is an act of will or a volition, and the relationship between it and the relevant bodily behavior is a causal relation.
The view that a human action is a piece of bodily behavior caused by an antecedent act of will or volition is a view which was extremely popular in eighteenth, nine teenth, and twentieth-century philosophy, and is still held
1C even today. One version of this view is to be discovered in the seventeenth-century philosophers Descartes and
Thomas Hobbes. Likewise, it is evident that John Locke held such a view.
A. Prichard, "Acting, Willing, Desiring," in The Nature of Human Action, op. cit., pp. 41-49. 35
Enough difficulties, however, with the volitional
theory in general, have been pointed out by several
philosophers so as to cast a great deal of doubt on it
as a viable solution to the problem of distinguishing
human actions and (mere) bodily movements. For, in the 17 first place, several philosophers have expressed grave
doubts about whether anyone ever does experience a voli
tion or act of will. Further evidence to support this
objection comes from the fact that, in ordinary discourse,
we seldom, if ever, speak of volitions of acts of will.
In any case, we seldom speak of such things as volitions,
and when we have occasion to speak of an act of will it
is usually in a context of great effort being required to
perform the task and especially in those situations in which there is also a strong tendency to do the opposite.
This suggests, of course, that Gilbert Ryle may have been
correct when he claimed that volitions and acts of will were merely theoretical notions invented by philosophers
to explain the alleged difference between human action and
(mere) movements of his body.
But no matter how valid the claim not to experience volitions is, it is not sufficient by itself to destroy
1>7See, for instance, Richard Taylor, Action and Purpose, op. cit. , p. 68? A. I. Melden, "Action," Readings in the 36 the volitional theory. For, first, it would be quite a simple matter for the defender of the volitional theory to claim that, contrary -to the testimony of Gilbert Ryle,
Richard Taylor, and A.I. Melden, they, at least, do exper ience volitions. As might be expected, the defender of the volitional theory would then most likely go on to add the further provision that those who claim not to experience volitions really do, but fail to correctly recognize them for what they are. This is the sort of claim which seems to be inherent in Myles Brand's criticism that Richard
Taylor and the others who employ this type of argument against the volitional theory have falsely supposed that if someone experiences something, then no arguments are neces sary to convince him that he does experience it.**-® This supposition is mistaken, Brand argues, because it is per fectly possible that a person can experience something with out being aware that he is in fact experiencing it. But in what sense or senses is it possible for an individual to experience something without being aware that he is exper iencing it?
Theory of Action, ed. by Norman S. Care and Charles Landesman, op. cit., p. 33; and Gilbert Ryle, The Concept of Mind (New York: Barnes and Noble, Inc., 1949), pp. 64-65. 1 8 Myles Brand, The Nature of Human Action, op. cit., p. 10. 37 The difficulty in answering this question arises from the fact that the word "experience" is itself a terribly vague word. In one sense of the word, to experience some thing is merely to live through the time at which it occurs. This is the sense in which some people, infants, for example, may be said to have experienced the Great
Depression. This sense of experience does not seem to carry the implication that the person is aware of what he is experiencing. However, there seems to be another sense of experience in which awareness of what is being experienced is a necessary condition. The sense in which a person could be said to be experiencing red must surely be a sense in which awareness is implied, and in this sense, Brand's claim seems self-contradictory. How could an individual be directly experiencing red and not be aware that he was?
And I can attach no significance to the notion of an indirect experience of red. If Brand's claim is going to make any sense at all, it must be interpreted in the first sense.
Now I think it can be argued that there are at least two kinds of cases in which a person may be said to be ex periencing something without being aware of what he is ex periencing. The first sense in which we may speak of a man's experiencing something but not being aware that he is, is 38 the sense in which we speak of unconscious experiences.
Hence, a mother who desperately wants to believe that her son is not a coward may hear the testimony which confirms this fact, but suppress it so that she is no longer con sciously aware of it. Strictly speaking, however, this is not a case of experiencing something and not being aware of experiencing it, but rather a case of experiencing something and suppressing its memory. A second sense in which we might speak of a man experiencing something without being aware that he is experiencing it, is the sense in which a man sees or hears something but he does not know what it is that he is seeing or hearing. This sense of •'experience" is spoken of by Roderick Chisholm in making the distinction be tween propositional and nonpropositional knowledge. He used the following quotation from Robinson Crusoe to illustrate the difference:
When, one morning the day broke, and all unexpectedly before their eyes a ship stood, what it was was evi dent at a glance . . . But how was it with Friday? As younger and uncivilized, his eyes were presumably better than those of his master. That is, Friday saw the ship really the best of the two; and yet he could hardly be said to see it at all.1^
It is possibly just this nonpropositional sense of "see" or
"aware" that Brand has in mind here. It is the sense in which
19Roderick M. Chisholm, Theory of Knowledge (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966), pp. 10-11. 39 a person may experience something but fail to identify or classify it properly. Hence, it is unquestionably true that Friday had approximately the same visual sensations that Crusoe had. All that can be said is that he did not know that what he saw was a ship, i.e., he failed to identify it as a ship.
But is it plausible to suppose that so many apparently intelligent and competent men could have failed so miserably to identify what they had experienced? Of course, the voli tional theorist may claim that everyone has volitions but many fail to recognize them simply because they do not know what to look for. But, in that case, the burden of showing that there are volitions falls on the volitional theorist, and so far as I know no volitional theorist has succeeded in clearly specifying what he means by a volition. That such a prospect is implausible is also reflected in the fact that, as A. I. Melden and others have pointed out, there are some very serious difficulties involved in attempting to specify criteria for identifying volitions. Hence, although Profes sors Ryle, Taylor, and Melden may not have refuted the voli tional theory by appealing to the lack of introspective evidence, nevertheless, by pointing out that men sometimes do act impulsively without any antecedent deliberation, 40 choice, or mental effort, they have cast a great deal of doubt on the view that volitions are either causally suf ficient or causally necessary for human action.
There are, however, much more serious difficulties connected with the causal volitional theory besides those associated with the apparent lack of introspective evidence for the existence of volitions. The most serious of these difficulties is the problem of correctly identifying and describing volitions themselves. This difficulty has been 20 ably pointed out by A. I. Melden. Let us grant for the moment that the causal volitional theory is true, i.e., that there are (mental) volitions which cause certain mus- cle movements to occur. 21 we may assume then that there will be a different and distinct volition for each different bodily movement. Hence, there will be a volition which causes my right bicep to contract, another volition which causes my left bicep to contract, still another volition for contracting my right thigh muscle, and so on. But if there is a different and distinct volition for each one of these
20 A. I. Melden, "Willing," in The Philosophy of Action, ed. by Alan R. White, op. cit., pp. 70-78. 21 Melden argues that the same difficulty would arise for any causal theory of human action regardless of whether the antecedent cause event is conceived of as a volition, choice, decision, etc., and regardless of whether this event is con ceived of as being mental of physical. 41 respective muscle movements, then we ought to be able to distinguish one from the other and describe that difference. Of course, it is not being claimed that each and every individual actually be able to perform this task of describing the relevant difference between these volitions, but only that it be theoretically pos sible for an individual to do so, and Melden seems to be arguing that it is not even theoretically possible to do this.
Nothing much is gained by suggesting that the rele vant volition may be described as the thing which causes the relevant muscle movement, e.g., "the volition which causes the contracting of my right bicep." But such an attempt is illuminating in that it implies that there is only one way in which we can give a differentiating descrip tion of a volition, viz., in terms of the corresponding muscle movement. That is, if Melden is correct, the only way in which we can identify and describe an act of voli tion is in terms of the corresponding muscle movement it is alleged to be the cause of. But if this is so, then no causal volitional theory can be true. The reason for this is that it has become a well-accepted doctrine in philosophy that whenever two things are causally related, the cause must be capable of being described independently 42
of the description of the effect. Hence, if volitions
cannot be described independently of the muscle move
ments they are alleged to cause, then it makes no sense
to claim, as this version of the volitional theory does,
that there are (mental) volitions which cause some muscle
movements to occur.
Melden believes that our inability to describe the
relevant difference between these various volitions except
in terms of the corresponding muscle movements is not due to an impoverished state of our natural language. It is not just a matter of not having at our command the neces sary words to do it with because even if we were to invent some new words to mark the differences, we would still have the problem of giving these new words meanings precise enough so that we could determine when they were being correctly applied and when they were not.
This argument of Melden's is a version of what has 22 come to be called "The Logical Connection Argument," and forms the basis of his and many other action theorist's objections to any causal account of human actions. I will not discuss the adequacy of this argument here since it will be thoroughly discussed in Chapter III. There I shall argue
22C. Raziel Abelson, "Doing, Causing, and Causing to Do," The Journal of Philosophy, LXVI (1969) p. 180. 43
that Melden’s argument is not sufficient to refute any
causal account of human actions, and that it gains what
ever plausibility it has solely from the fact that it
confuses the notion of an intentional object with the
intention itself.
In all fairness to the causal volitional view, it
should be pointed out that the claim that it leads to an 23 infinite regress is fallacious. This can be clearly
seen by examining the way in which Melden has argued for
this claim. Hence, he argues,
Grant for a moment that an event labelled "an act of volition” produces a muscle movement; there is a difference surely between the occur rence of such an event and my producing it. We saw that there is a difference between the occur rence of a muscle movement and my moving that muscle; hence it was that the supposition of acts of volition was invoked. But equally there would seem to be a difference between the occurrence of an act of volition and my performing such an act. Who can say that volitions may not occur through no doing of the subject and in consequence of interior mental events deep within the hidden recesses of the self? If so, willing the muscle movement is not enough; one must will the willing of the muscle movement, and so on ad infinitum.
This argument, however, is clearly fallacious. For it mistakenly supposes that the causal volitional theorist must
hold that volitions are themselves actions. It is easy to
2^Gilbert Ryle, The Concept of Mind, op. cit., p. 69; and A.I. Melden, "Willing," The Philosophy of Action, ed. by Alan R. White, op. cit., p. 71.
^Ibid., p. 71. (Italics mine.) 44
see how this assumption could explain why Melden believes
that the theory leads to an infinite regress. For, if
raising my arm is an action if and only if it is caused by
a volition and that volition is itself an action, then it
will have to be caused by other volitions, and so on ad
infinitum. But, of course, there is no reason why the
causal volitional theorists cannot claim that at least some
volitions are not actions, and still claim that the dif
ference between bodily actions and (mere) bodily movements
is that the former are caused by volitions while the latter
are not. But Melden is not alone in making this unwarranted
assumption. It can also be seen in the following argument
of Gilbert Ryle:
So what of volitions themselves? Are they voluntary or involuntary acts of mind? Clearly either answer leads to absurdities. If I cannot help willing to pull the trigger, it would be absurd to describe my pulling it as "voluntary." But if my volition to pull the trigger is voluntary, in the sense assumed by the theory, then it must issue from a prior volition and that from another ad infinitum.25
At least part of the plausibility of Melden's argument
springs from his insistence that there must be a difference
^Gilbert Ryle, The Concept of Mind, op. cit., p. 67. (First italics mine.) An interesting aspect of Ryle's argu ment here is that it seems to be assuming that nothing is to be considered an action of mine unless it v/as caused by a voluntary volition. But in spite of the grammatical simi larity of "voluntary" and "volition," there is no reason, as we have just seen, why the causal volitional theorist need claim that volitions themselves are either voluntary or actions. 45
between a volition just occurring and the agent’s pro
ducing one. Why this must be the case is a bit of a
mystery, 'however. The only plausible suggestion is that
Melden is working under the supposition that even if actions
are produced by some causal antecedents, those antecedents must themselves be performed by the agent if they are to
be considered actions. However, we have already seen, in
conjunction with Richard Taylor's arguments to the effect
that action sentences are not capable of being analyzed nomically, that this is an unwarranted assumption.2®
Even though none of these difficulties with the causal volitional theory have proved to be conclusive they do seem to suggest that the causal volitional theory gener ates more problems than it succeeds in solving. Volitions are seldom mentioned in our ordinary discourse about what we and others do, and they seem to be used primarily for cases in which what is done involves some real possibility of failure. But while this fact about ordinary usage has not gone unnoticed, it is diffucult to find any other posi tive arguments against the view that raising an arm is merely the effect of a volition. However, one argument is to be found in Arthur Danto's article, "Basic Actions."
2®See pp. 22-24 of this paper. 46
Danto argues,
A paralytic might think there is some effort he is not putting forth, by which, if he did or could put it forth, he might as a consequence move his arm. But I want to say that he cannot try to move his arm if moving his arm is not already in his repertoire of basic actions. So in a sense he is right. If he could make the required effort, he could move his arm. But he cannot in the only appropriate sense move his arm. Consider the analogous situation with someone epistemically abnormal, say a deaf man. To ask a deaf man to try to hear a certain sound is ren dered inappropriate by the fact that he is deaf. To try to hear, say, faint and distant music is to make an effortful listening. Only those who can already hear can make this effort. And what would count as trying (listening) in the deaf man's case? He could cup his ear, could place his ear to the ground, could contort his face and close his eyes. All this, however, is the pantomime of listening. Had he grinned or wagged a finger, it would have been as helpful. For there is no one thing that is better than any other in his situation. It is exactly this way with trying to move an arm. It is appropriate only to ask someone to try to move his arm when something externally inhibits normal movement, e.g., the arms are pinioned, and cannot be moved freely and without effort. But the para lytic cannot move his arm at all T . . One can do with effort only what one can do effortlessly? and trying, the effort of will, is not something apart from the action that stands to it as cause to effect. It is the required action already being performed in untoward circumstances. Doing something with effort is not doing two things, any more than doing something gracefully is doing two things. Moving an arm is not then the result of an act of will? it is an act of will. But to speak of an act of will wKen the going is smooth is to behave a little like the dypsomaniac who wants to know what sorts of pink rats ordinary people see.27
27Arthur Danto, "Basic Actions," in The Nature of Human Action, ed. by Myles Brand, op. cit., pp. 264-265. 47 ( Dantb’s argument seems plausible enough, but there are at least two replies that the volitional theorist might make against it. First, he could attack the supposed iden tity of willing and trying, claiming perhaps that although willing is neither an action nor sufficient to cause the intended action, trying is an action and some willing is causally necessary to produce it. Thus, in the case of paralytics and deaf men, there may be willing but no trying.
In the second case, he could grant the identity of willing and trying, but claim that neither is something a person does. The willings (tryings) which occur in normal per sons simply do not occur in the case of the paralytic or deaf man. Both of these rebuttals seem to be in the same spirit as the volitionalist's replies to the infinite regress argument.
Barring Melden's objection that there is still the difference between my willing something and a willing simply Q occurring in me to be accounted for, I think at least two things can be said about these two attempts at defending the volitional theory. First, as against the second rebuttal, if willing is identified with trying and these are thought
2®A. I. Melden, "Willing," The Philosophy of Action, ed. by Alan R. White, op. cit. 48 of as mental events which need to have any physical effects at all, then it is extremely difficult to see how trying to perform a required movement can be significantly dis tinguished from merely concentrating on, wishing for, or vividly imagining the required movement. A paralytic may very well concentrate on moving his arm, wish for his arm to move, or even vividly imagine his arm moving, but none of these is the same thing as actually trying to move his arm. Likewise, it would be extremely odd to say that a man was trying to open a door or trying to write an essay, but at the same time simply sitting in his armchair doing nothing, but just mentally trying to open the door or to write an essay. Hence, with regard to the performance of some observable overt bodily movements it is false that an agent is trying to do it unless his trying to do it involves the exertion of some actual physical effort or other, i.e., unless it involves the actual observable tensing of some bodily muscle, or the occurrence of some physiological or 2 9 neurological event or other.
This point is closely related to another feature of the notion of trying, namely, we do not, except in fairy tales or other works of fiction, speak of a man as trying
29Cf. Richard Taylor, Action and Purpose, op. cit., pp. 79-85. 49
to do something unless he possesses the necessary ap paratus (e.g., muscles, sense organs, nerves, etc.) to do it with. Thus, we do not ask human beings to wiggle a single hair on their heads, or to try to wiggle a fingernail independently or their fingers in contrast to merely concentrating on, wishing for, or vividly imagin ing these things because human beings simply lack the suitable apparatus necessary for doing so.
The second thing that can be said about these at tempts to escape Danto's argument is that although they show that a causal theory of human actions is possible, the volitional theory does seem to be out of step with the way in which we ordinarily speak about and understand actions. A better theory would be one which talked about reasons (e.g., wants, desires, purposes, intentions, etc.) as the causes of actions. And since there seems to be no independent evidence for the existence of such things as volitions nor any great demand for them in the way in which we ordinarily speak about actions, volitions appear to be nothing less than theoretical fictions invented by philo sophers to explain the alleged difference between actions and movements. 50
2.7. Ascripticism
Faced with the inadequacies of the volitional theory/
some philosophers have suggested that a human action is a
piece of behavior to which responsibility can be ascribed.
Thus, in his influential article, "The Ascription of
Responsibility and Right," H.L.A. Hart rejected what he
called "the old-fashioned" and "the modern" answers to the question, "What is a human action?"3^ "The old-fashioned
answer" is the view that human actions can be analyzed in terms of some piece of bodily behavior plus some antecedent mental event, and "the modern answer" is that a complete theory of human action can be constructed in terms of bodily behavior plus dispositions to behave in certain ways. The common error of both of these answers, Hart believed, was that they mistakenly supposed that some combination of purely descriptive sentences could amount to a correct analy sis of the concept of a human action. This error resulted,
Hart argued, from the failure to realize that the primary function of our discourse about human actions is not to describe them but rather to ascribe responsibility.31 in arguing for this, Hart compared the concepts of action with
^ h .L.A. Hart, "The Ascription of Responsibility and Rights," Logic and Language (First Series), ed. by Antony Flew (Oxford: Basil Blackwell, 1955), pp. 145-166.
31Ibid., p. 161. 51
the legal concept of rights, and concluded from the fact
that the concept of rights was defeasible, i.e., claims
about rights can be overridden or at least weakened by excuses and mediating circumstances, and therefore are not merely descriptive but also ascriptive, that the concepts of action were also not merely descriptive, but ascriptive as well.^^ Furthermore, Hart seems to be of the opinion that action sentences can only be used descriptively because there exist certain conventions of responsibility.
The importance" o"f Ascrxptivism for the theory of action seems to be analogous to the importance of Emotivism for ethics. For just as the emotivist in ethics wishes to stress the notion that the primary function of ethical utterances is not to describe the actual features of situa tions , but rather to perform some other function such as express emotions, commend, etc., the ascriptionist wishes to stress the notion that utterances about human actions have as their primary function the ascribing of responsibility rather than the describing of the actual features of situa tions. And just as the truth of Emotivism would render useless any attempt at defining such notions as right, good, ought, etc., the truth of Ascriptivism would rule out any attempt at defining human actions.
32Ibid., p. 160. 52
In any case it is clear that the tenability of As-
criptivism depends to a large extent on the ascriptivist's
ability to draw a clear-cut distinction between ascribing
on the one hand, and describing on the other. Unfortunately,
so far as I can determine, Hart himself never makes a sub
stantial effort at explaining this crucial distinction.
Sprinkled here and there in his original article one can
discover a few clues, but there is hardly anything which
could be termed a thorough explanation of this distinction.
Thus, in speaking about the concept of action, Hart writes,
It is fundamentally not descriptive but ascriptive in character; and it is a defeasible concept to be defined through exceptions and not by a set of necessary and sufficient conditions whether physical or psychological.
And in another place it looks very much as if Hart wants descriptive concepts to be viewed as concepts whose presence in sentences accounts for the possibility of truth and falsity, whereas ascriptive concepts do not have this feature. Hence, given the supposed parallel between the concepts of action and the legal concept of rights, Hart claims,
First, we, of course, very often make use of legal concepts in descriptive and other sentences and the sentences in which we so use them may be state ments and hence (unlike the judge's decision in which legal concepts are primarily used) they may be true or false.34
■*^Ibid., pp, 161-162. 34 Ibid., p . 156. 53
This feature certainly suggests that Hart's ascriptive-
descriptive distinction may be the same as the better-
known prescriptive-descriptive distinction.
It has become customary among many philosophers to
distinguish between "descriptive" meaning and what they 35 qc have labeled "emotive" or "evaluative"Jb meaning. This
dicotomy is in fact so prevalent in contemporary philosophy
that one can hardly pick up a recent work on any philoso
phical problem without discovering at least the rudiments
of such a distinction there.3? But in spite of its popu
larity, it is extremely obscure how the distinction between
descriptive and emotive or evaluative meaning is to be
interpreted, and it is by no means always clear exactly what
it is that a philosopher means to be contrasting when he
contrasts these two kinds of meaning. Descriptive sentences,
as opposed to prescriptive or emotive sentences, are sup
posedly those sentences which are capable of being true or
35 C.L. Stevenson, Ethics and Language (New Haven: Yale University Press, 1944), p. 59. 3 6 R.M. Hare, The Language of Morals (Oxford, England: Oxford University Press, 1952), Chapter VII. 37 The descriptive-prescriptive distinction has also spread to talk about kinds of laws, and hence, it is not unusual to discover philosophers who attempt to distinguish between what they call "descriptive" and "prescriptive" laws. Cf. Moritz Schlick, Problems of Ethics, Chapter VII in 54
false. A not too uncommon way of expressing this has been
to say that whereas descriptive sentences are or express
statements, prescriptive or emotive sentences do not. But to say this is totally uninformative unless we know pre cisely what is being asserted when a sentence is said to be capable of being either true or false.
One way of rectifying this situation would be to attempt to specify exactly what kind of truth or falsity is being talked about. Thus, for instance, it is not too difficult to find philosophers claiming that a sentence is, or expresses, a statement if and only if it is capable of being objectively true or false. But exactly what is this all-important word "objectively" supposed to do here? The best, and sometimes the only, answer that a theorist who insists on a sharp dicotomy between descriptive and pre scriptive meaning gives to this question is that a sentence is capable of being objectively true or false if and only if it asserts or can be interpreted as asserting something about the world, i.e., that the sentence asserts or denies
A Modern Introduction to Philosophy, ed. by Paul Edwards and Arthur Pap (New York: The Free Press of Glencoe, 1957) , p. 350. John Hospers, An Introduction to Philosophical Analy sis, Second Edition (Englewood Cliffs, New Jersey: Prentlce- Hall, Inc., 1967), pp. 230-231. 55
the existence or non-existence of some fact, state of
affairs, situation, or set of facts, states of affairs, or
situations. Yet, this answer is as uninformative as the
previous one unless some way of telling whether or not a
sentence asserts or denies something about the world can
be given.
In spite of the fact that many recent philosophers
have realized that the difference between describing something
and evaluating something, expressing something, and so on,
is a distinction between the ways in which sentences and
words are used, enormous difficulties seem to stand in the
way of providing explicit analyses of these different uses.
J.L. Austin, for instance, attempted to explore this rela tively unexplored territory in his book How to Do Things With 3 8 Words. As Austin correctly noted not only to we do things by moving our hands, fingers, legs, etc., we frequently do things by uttering certain words (e.g., we promise, insult, bet, convince, and so on.) The doing of these things with words is quite different from merely uttering the words, although in certain contexts merely uttering the words "I promise" might be sufficient for performing the act of
■^J.L. Austin (Cambridge, Massachusetts: Harvard Univ. Press, 1962). 56
promising. Austin labeled such utterances performative
utterances and claimed that such utterances were more
appropriately spoken of as being either "happy" or"unhappy" 3 Q rather than true or false. In making the contrast between
reporting that a promise had been made, a request made, or
a comment made, and actually promising, requesting, or
commenting, Austin had apparently intended to be drawing
some distinction between "statements" and performatives,
i.e., between uses of sentences which were capable of being
either true or false, and uses of sentences which were not
capable of being either true or false. His hopes, however,
fizzled when he came to the realization that some performa
tives, what he called expositives (e.g., stating and main
taining) , showed signs of being true or false:
. . . beginning with the supposed contrast between performative and constative utterances, we found sufficient indications that unhappiness nevertheless seems to characterize both kinds of utterance, not merely the performative; and that the requirement of conforming or bearing some relation to the facts, different in different cases, seems to characterize performatives, in addition to the requirement that they should be happy, similarly to the way which is characteristic of supposed constatives . . . we still have utterances beginning 11 state that ...' which seem to satisfy the requirements of being performative, yet which surely are the making of statements, and surely are essentially true or false.40
39j .l . Austin, How to Do Things With Words (Cambridge, Massachusetts: Harvard Univ. Press, 1962), pp7 12-14.
40Ibid., p. 91. Faced with this result, Austin made a new start and
distinguished three different kinds of acts which could
be performed with v/ords; the result was his now famous
distinction between locutionary, illocutionary, and per-
locutionary speech acts. But, even armed with this new
set of distinctions, Austin found himself unable to arrive
at any clear-cut, rock-bottom distinction between the
descriptive use of sentences and expressions, and other
uses of sentences and expressions.4 Apparently, Austin
thought that there was at least one major difficulty which
stood in the way of any attempt to analyze the illocutionary
act of making a true or false statement, and thereby, distinguishing it from the performance of other illocuationary
acts. The difficulty involved the possibility of finding
any simply answer to the question of whether or not a par
ticular sentence or expression was on a particular occasion used to state something either truly or falsely, and Austin
seemed to be arguing that there just is not any simple man ner in which this question could be resolved. Consider, for example, "France is hexagonal." Has what has been said been stated truly or falsely? Now, apparently, Austin
4^Ibid., pp. 140-145. 58
thought that, ordinarily at least, there was no point in
asking such a question independently of the specific
purposes and intentions of the person making this utterance
and his audience, our knowledge of the facts, and the
specific circumstances surrounding that utterance. Austin
did admit, however, that we may aim at some kind of ideal
of what it would be like for an utterance to succeed in making a true statement in all circumstances, for any purpose, to any audience, and so on, but on the whole, he seems to be rather skeptical of the possibility of judging
some statement or description true of false except on the basis of some specific purposes and intentions, circum stances, and relative knowledge of the facts. Hence, if
Austin is correct in this, and I believe he is, then it would appear that the prospect of providing any general analysis of the illocutionary act of making either a true or false statement, or of giving either a true or false descrip tion of something, is extremely dim. Additional evidence for this comes from the fact that, so far at least, no such analysis has appeared.
The prescriptive-descriptive distinction owes a great deal of its plausibility to the labors of R.M. Hare in his 59
books, The Language of Morals,4^ and Freedom and Reason.4^
Yet one can search these works in vain for any clear and
precise formulation of the difference between prescriptive meaning and descriptive meaning. About as close as Hare
comes to precisely characterizing descriptive meaning is to say that a term has descriptive meaning provided that there is some rule attaching the term to a certain kind of object.44
The basic correctness of this can, I believe, be seen by considering the following relevant meanings of "descrip tion" given by The Oxford English Dictionary: (1) "The action of setting forth in words by mentioning recognizable features or characteristic marks; verbal representation or portraiture." (2) "A statement which describes, sets forth, or portrays; a graphic or detailed account of a person, thing, scene, etc..." (3) "The combination of qualities or features that marks out or serves to describe a particular
42r .m . Hare, The Language of Morals, op. cit.
4^R.M. Hare, Freedom and Reason (London, England: Ox ford University Press, 1963).
44Ibid., pp. 7-14. So far as I can determine Hare never attempted to formulate the defining characteristics of pre scriptive language in general, except to say that prescriptive language can be divided into two major kinds, viz., impera tives and value judgments. Cf. R.M. Hare, The Language of Morals, op. cit., p. 3ff. 60
class. Hence, b. A sort, species, kind, or variety,
capable of being so described.* 4 JS Definition three clearly
suggests that to describe something is really to classify
it in a certain fashion, and it is not too difficult to
find philosophers who do identify the process of describing
with the process of classifying.
But classification and description are really the same process. To describe a given animal as carnivorous is to classify it as a carnivore; to classify it as a reptile is to describe it as reptilian. To describe any object as having a certain property is to classify it as a member of the class of objects having that property.46
Definition one, however, may appear to be emphasizing
something else, viz., the use of words which correspond to recognizable features or characteristic marks of the thing being described without regard to how these things are
classified. But, the sense of the word "describe" in which to describe something is to classify it also seems to imply the ability to recognize and distinguish various things according to their features or properties. Thus, for instance,
46The Oxford English Dictionary, Vol. Ill, p. 238.
46Irving Copi, Introduction to Logic, (2nd ed., New York: The Macmillan Company, 1961), p. 461. to classify something as a reptile_seems to imply the
ability to recognize the characteristics of reptiles and
also the ability to distinguish these features from the
features of mammals, fish, birds, etc. Hence, it is easy
to suppose that the ability to describe something and the
ability to understand a description likewise depend upon having the ability to recognize and distinguish between the various features or properties of things. In view of this, it is not difficult to see how questions about how various symbols and sentences can be used to describe some thing can get translated into epistemological queries, i.e., into questions about how we know that a term is being cor rectly applied to a situation or not.
There is, of course, a world of difference between correctly describing something and incorrectly describing it. In general, to correctly describe something is to men tion those features or characteristics that the thing in question actually has, while to misdescribe something is to mention features or characteristics that it does not possess.
Misdescriptions are not the nonexistence of descriptions, but the use of symbols or sentences which fail to pick out the actual features or characteristics of the thing being described. And the difference betv/een a correct description 62 and an incorrect description of something depends, in part, on the difference between the correct application of terms and the incorrect application of terms. But what is the difference between correctly applying a term and incorrectly applying a term? One way of putting this question is to ask, "How can we know whether or not a specific linguistic expression fits a given object, state of affairs, or situa tion?" This question is one with which Wittgenstein did a great deal of struggling, and his denial of the possibility of a private language in the Philosophical Investigations depended, to a large extent, on the answer that he finally came to. And if I understand his answer correctly, the question of whether or not a word is being correctly applied or not must always be a "public" question. No private sensation or. image could fulfill the necessary requirement.
We discover whether or not a linguistic expression correctly applies to a fact, state of affairs, situation, or object by observing the characteristics or properties of that fact, state of affairs, situation, or object. Thus, the question of whether or not I have used the word "red" correctly in describing my copy of Free Action is a question to be re solved by publicly observable characteristics of that book, e.g., by pointing to its publicly observable color. This answer seems to me to be substantially correct
and I do not wish to quibble over it. But, instead I would like to question a certain conclusion which many philosophers have thought follows from it, namely, the view that there are certain epistemological atoms in virtue of which the question of whether or not a particular lin guistic expression has been correctly applied or a' parti cular description is correct can be completely determined.
All classical empiricists, from Locke to Russell, and the positivists begin with the thesis that all genuine know ledge of the world depends on observation, and what they wish to establish is some theory of what is observed which would free that knowledge from any elements of interpretation or appraisal. That is, they assumed that there must be some class of entities or properties which v/ould provide the ultimate basis of evidence, and that, thus, some ultimate distinction must be made between what is observed and what is merely inferred. Hume was thus led to construct a theory of empirical knowledge in which observations are of simple impressions, which were incorrigible and unanalysable episte mological atoms. The positivists, in turn, were led to deny cognitive significance to every sentence except those which were in principle verifiable, and Carnap was to construct 64 his system in Per Logische Aufbau Per Welt out of "protocol- sentences" which were as close to epistemological primary elements as possible.
But the view that there is necessarily one and only one correct way in which the world is to be apprehended and described has been severely criticized by a number of contemporary philosophers.^ Hence, it v/ould appear that the supposed distinction betv/een descriptive and prescrip tive meaning is not as clear as some of its proponents seem to think it is, and until the proponents of this distinction are able to provide a clear-cut and non-controversial account of the distinction, the tenability of any theory such as
Ascriptivism v/hich is based upon it is bound to remain equally debatable.
In addition to these difficulties, George Pitcher has criticised Hart on the grounds that, structly speaking, a person is responsible for the consequences of his action, not for the action itself.Suppose, for example, that
47 Cf. Nelson Goodman, "The Way the World Is," Review of Metaphysics, XIV (1960), pp. 48-56; Ivor Hunt, "Value-Con- cepts and Conceptual Truth," Proceedings of The Aristotleian Society, LXIII (1962), pp. 33-44; A.R. Louch, Explanation and Human Action, (Berkely and Los Angeles, Calif.: Univ. of Calif. Press, 1966), Chap. Four; S.E. Toulmin and K. Baier, "On Describing," in Philosophy and Ordinary Language, ed. by Charles E. Caton, (Urbana, 111.: Univ. of 111. Press, 1963), pp. 194-219.
48Qeorge Pitcher, "Hart on Action and Responsibility," in Philosophical Review, LXIX (1960), pp. 226-235. 65
Dodd is mixing cocktails for his guests. But in doing so he is insufficiently cautious and unknown to him spills some of the ingredients on the floor. This then causes one of his guests to fall and injure himself.
Here, it would appear that Dodd is responsible for the dangerous situation and the resulting injury to his guest, although spilling the ingredients on the floor was not an action of his since it was nothing he did intentionally.
This criticism, however, is misleading. For it is one thing to argue that a person may be held responsible for both the intentional and unintentional consequences of his actions, and quite another thing to argue that a person is only responsible for the consequences of his actions and not the actions themselves. The view that a person may be held responsible for the consequences of his actions may very well be consistent with Hart's view. For it has been argued that there is absolutely nothing to prevent us in some cases from expanding the description of a person's action to include its consequences as well as the action itself. This feature of our language whereby what an agent does may be described either as "telling a lie" or as "telling a life-saving lie" makes it possible to pack the consequences of an action into the very description of the action itself.
Thus, in this regard, the view that a person is only respon sible for the consequences of his action but not the action 66
itself is not defensible, and Hart's view is consistent with a person's being responsible for the consequences of his actions.
But in spite of its initial plausibility I believe that it can be shown that Hart's analysis of action sentences as primarily ascriptive rather than descriptive is mistaken. First, although I do not intend to get deeply involved in the nuances of excusing conditions, it seems clear that Hart's case for the defeasibility of action concepts places too much emphasis on legal verdicts (e.g.,
"Smith is guilty of murder") and special cases (e.g., con tracts). Hence, Hart claims about the judge's verdict that
. . . since the judge is literally deciding that on the facts before him a contract does or does not. exist, and to do this is neither to describe the facts nor to make inductive or deductive inferences from the statement of facts, what he does may be either a right or a wrong decision or a good or bad judgment and can be either affirmed or reversed and (where he has no jurisdiction to decide the question) may be quashed or discharged. What cannot be said of it is that it is either true or false, logically necessary or absurd.49
49H.L.A. Hart, The Ascription of Responsibility and Rights, op. cit., p. 155. 67
But regardless of how valid this observation is with regard
to special cases like murder, rape, trespassing, etc., it does not apply to all judgments to the effect that "S did it" where the "it" is anything one pleases, e.g., killed, copulated, stepped over the fence, etc. And rather than
serving the purpose of ascribing responsibility, these latter action sentences clearly are capable of being true or false. So, unless Hart wishes to defend the untenable thesis that no judgment based on evidence is capable of being either true or false, it is simply false that all sentences containing action concepts have as their role the ascription of responsibility.
Hart's analysis can be criticized on other grounds as well. For if what we are ascribing to a person in saying that he performed a certain action is responsibility, then we are faced with the problem of exactly determining what is involved in claiming that a person is responsible for something. But the notion of responsibility is, unfortunatley, quite vague, and perhaps, even more unclear than the notion of an action itself. In his article "Responsibilities and
Action,"50for instance, Kurt Baier distinguishes at least
^°Kurt Baier, "Responsibilities and Action," The Nature of Human Action, ed. by Myles Brand, op. cit., pp. 105-107. five different senses of "responsibility" within the
context of bringing people to account. The first sense is accountability. In this sense a person is said to be responsible for his actions because he has all of the necessary and sufficient psychological faculties for having the ability to be guided by certain social rules and practices. Children and madmen are assumed not ac countable for their actions in this sense. Ascribing responsibility to a person may alse mean holding him responsible for performing certain specific tasks such as feeding the dog. The third sense of "responsibility" is answerability. This is the sense in which we may claim that an agent is responsible for failing to satisfy some socially obligatory requirements such as a captain being held answerable for his ship going down with great loss of property and lives. The fourth sense of "responsibility" that Baier distinguishes is culpability. This is the same as answerability except in this case the person has no adequate excuse for failing to satisfy the socially obliga tory requirements as when the ship’s going down is a result of the captain's negligence. And finally, ascribing re sponsibility to an individual may be holding him liable for his culpable actions in the sense that we are prepared either 69
to punish him, to condemn him, or to force him to com
pensate in some way for his actions.
These five different senses of responsibility form the
basis of Baier's criticism of Hart's ascriptive analysis.
If I understand him correctly, his criticisms are these:
(1) People frequently do things for which there are no
contrary social rules or practices, e.g., closing the door.
In this case, there cannot be any question of ascribing responsibility to them in any one of the five senses that
Baier has distinguished. (2) People can be held responsible
in the sense of being answerable for the actions of others although the other is not necessarily responsible for his actions, e.g., I may be held answerable for the damage that my minor son does to my neighbor's window, although he may not be held responsible for it. (3) Someone may do something which he is required not to do and still not be held responsible for it, because he is not accountable for any of his actions. And (4), to say, "He did not do it" may also succeed in ascribing responsibility.
These criticisms are, I believe, serious enough to mitigate much of the original plausibility of Hart's account of ascripticism. But they have not appeared so to others.
51Ibid., pp. 112-113. 70
Joel Feinberg, for instance, has argued most convincingly
(but to my mind unsuccessfully) that Hart's notions of
ascriptiveness and defeasibility can be successfully ex
tended so that they apply to all actions including those
that do not involve the failure of an agent to satisfy some
relevant social and/or moral rule or p r a c t i c e . Actually,
all that Feinberg does is merely stretch the notion of
responsibility so as to include much more than what is
usually included in that concept, and in so doing he has
left the door open to a causal analysis of actions. And it was exactly the supposed failure of such causal accounts, among other things, which led Hart to champion ascripticism
in the first place.
In his article "Action and Responsibility," Feinberg distinguished at least five different things that could be meant by the phrase "ascription of responsibility." The first of these is the straightforward description of caus ality, e.g., Smith pushed the button causing the door to open.
The second major sense of "Ascription of responsibility," ascriptions of causal-agency, requires that a preliminary distinction be made between simple and complex actions.
Simply stated, complex actions are those actions which re quire that the agent do something earlier as means for doing 71
it, e.g., performing the complex action of mowing your
lawn usually requires some necessary preliminary steps.
Simple actions, on the other hand, require no earlier
actions as means, e.g., smiling, frowning, raising one's
arm, and so on. Ascriptions of causal-agency only apply
to complex actions. Given that, we may describe a man's
action either so that it includes the "consequences,"
"results" or "effects" of his action (e.g., "Smith
startled Jones"), or so that it excludes these (e.g.,
"Smith clapped his hands."). This feature of our language
Feinberg labels "the accordion effect," and although I do not think that such a sweeping thesis can be adequately defended, as I will show in Chapter IV, its validity is
immaterial to my present purpose, which is simply to show that Feinberg's analysis of the concept of responsibility is too high a price to pay for saving ascripticism. It follows from the accordion effect that we can translate 4 any ascription of Smith's causality into an ascription of
Smith's causal agency. Thus, instead of saying "Smith clapped his hands causing Jones to be startled," we may say simply, "Smith startled Jones." The third type of ascriptions of responsibility are simple agency ascriptions, i.e., as cription of responsibility for performing simple actions, 72 for example/ "Smith raised his arm." Imputations of fault/ the fourth type, ascribe responsibility for a rather loosely- knit class of faulty or defective actions. The fifth re sponsibility is the kind of thing which might be meant by ascriptions of liability which, through a set of rules or practices, makes the agent liable for some sort of un wanted treatment.
Probably the most important consequence of this five fold classification is that the first three, straightfor ward causality, causal agency, and simple agency, are senses in v/hich responsibility is ascribed for actions v/hich are normal. That is, what Feinberg hopes to show with these different senses of ascriptions of responsibility is that responsibility is not alv/ays tied up with a context of moral and/or social rules and practices and can be ascribed to cases in which there are no such relevant rules or prac tices. Ascriptions of straightforward causality, causal agency, and simple agency are supposed to be genuine ascrip tions of responsibility even though there is no question of a failure to satisfy a relevant social and/or moral rule of practice, or of fault, desert of punishment, and so on.
No matter how successful Feinberg's five-fold clas sification of the different senses of the phrase "ascription 73 of responsibility" may be against Baier's criticisms, it is not without serious difficulties of its own. In stretching the notion of responsibility in this way it seems to me that Feinberg has left the door open to almost any kind of an analysis of action and action-terms, in cluding a causal analysis. For if ascriptions of straight forward causality, causal agency, and simple agency are all ascriptions of responsibility, it is difficult to see how this could be in conflict with what the volitional and dispositional theorists are claiming. To say, for instance, that Jones clapped his hands and thereby startled
Smith, or simply Jones startled Smith, or Jones raised his arm, is not to deny that a volition or a disposition to behave in a certain v/ay under certain circumstances had something to do with bringing these acts about. And if these are the senses in which we are ascribing responsibility in attributing these acts to Jones, it is highly questionable how volitional and dispositional accounts can be ruled out.
In short, on Feinberg's classification it is no longer clear how ascriptions of responsibility are to be contrasted with being caused by a volition or an activated disposition to behave. In addition to this, Feinberg's classification of descriptions of causality as "ascriptions of responsibility" seems to be totally unwarranted. For when a physicist at tributes the cause of an unusual amount of electromagnetic interference to an increased number of sunspots, in what sense can he be said to be "ascribing responsibility" to the sunspots? The only apparent sense in which he can be said to be ascribing responsibility to the sunspots is the sense in which he is asserting that the sunspots caused the increase in electromagnetic interference. But, in this case, this is no different from merely describing the sunspots as the cause of the increase in the electromagnetic interference. Surely, to claim, as Feinberg does, that attributing causes to things is really a way of ascribing responsibility to them is tanta mount to collapsing any distinction between ascribing and describing altogether. And in that case, ascripticism loses all of its explanatory power, since almost every statement could be construed as an ascription of responsibility.
Furthermore, Feinberg himself has claimed that as cribing and describing are two different uses of action sentences; the use depending on what the user wants to do with it. In fact, Feinberg believes that any kind of action sentence can be used either descriptively or ascriptively.
But even if we grant this, it does not follow that what 7 5 Feinberg labels ascriptions of simple agency such as "Smith
raised his arm" are not used to make statements which are
either true or false. For Feinberg also admits that such
sentences can be used either to identify the agent or to
identify the action itself, and such uses of these sentences
give every appearance of being factual uses. Nor are sen
tences attributing simple agency to persons, e.g., "Smith
raised his hand" always used to ascribe responsibility ex
cept in the trivial sense in which it says simply that Smith
was responsible for raising his hand. Hence, rather than
succeeding in extending the scope of ascripticism, Feinberg's
five different senses of "ascriptions of responsibility"
have made it indistinguishable from almost any other analy
sis of action sentences.
Ascripticism has also been refuted, I believe, on
completely different grounds. P.T. Geach has argued that
ascripticism mistakenly confuses predicating an action and c o calling something an action. According to Geach, ascrip
ticism mistakenly assumes that the primary use of the term
"action," along with other action-terms (e.g., "intentional,"
"done on purpose," "voluntary") is to call something an action.
53P.T. Geach, "Ascripticism," in Ethics, ed. by Judith J. Thomson and Gerald Dworkin (New York: Harper and Row, 1968), pp. 22-26. 76
This is a mistake, Geach argues, because not only is it the
case that these terms are used in predication, their
primary use is in predication.^ The evidence that action-
terms are used to predicate something of a subject is that
they may appear in conditional and disjunctive sentences,
alike. Thus, in the sentence, "If Jones' flipping the
switch is an action, then Jones is responsible for Smith's
death," the action is predicated of "Jones' flipping the
switch," but it has not been called or labelled an action
since the truth of the conditional sentence is consistent
with its not being an action, i.e., Jones' flipping the
switch may have been the result of an accidental movement
of his arm while he was in the process of swatting at a
fly. Borrowing from Frege, Geach claims that in order for
a sentence in which an action (or action-term) is predicated
of a thing to count as calling something an action, the
sentence must be used assertively, but this is not predi
cation. Hence, the primary use of action-terms is in pre dication, and not in calling something an action. If action-
terms did not have a predicative use in declarative sentences,
then that would mean that the inference
(1) If Jones' flipping the switch is an action, then he is responsible for Smith's death.
54Ibid., p. 25. 77
(2) Jones’ flipping the switch is an action.
(3) Therefore, he is responsible for Smith's death. would be invalid because it equivocates on the phrase
"Jones' flipping the switch is an action," whereas it is valid. Thus, if Geach is right here, and the primary use of action-terms is in predicating action-terms to some thing, then actions can only be explained by explaining what it means to predicate action-terms to something. But explaining what it means to predicate an action-term of a subject is the problem of explaining what an action, the problem to which ascripticism was purported to supply an answer.
Ascripticism is perhaps illuminating in the way in which it draws our attention to certain features of the concept of a human agent or person and a human action. Hart, for example,in stressing the idea that responsibility is intimately tied up with the existence of certain moral and social rules and practices has succeeded in bringing out one important feature of our concept of a human agent, viz., that a human agent is one who can in some sense be influ enced by these moral and social rules and practices. And seeing something as an action is frequently a matter of seeing it as an instance of certain moral, legal, and social rules or conventions. 78
2.8. Contextualism
In "About Behaviorism, "^Paul ziff appears to claim
that there is indeed a relevant difference between a human action on one hand and a (mere) bodily movement on
the other. For there is a difference, he claims, be tween someone gnashing his teeth and his teeth gnashing, and the.major difference between them, he suggests, is a C £ matter of certain "contextual and relational matters."
These contextual and relational matters are, for Ziff, sub sequent behavior. According to Ziff, one decides between the correctness of the expressions "S is gnashing his teeth" and "S's teeth are gnashing" on the basis of the consistency of inconsistency of S's subsequent behavior
(both verbal and nonverbal) with it. Actually, this needs some revision since, as Ziff claims it would be odd to assert that an individual discovers that he himself is gnashing his teeth rather than having his teeth gnash by attending to his subsequent behavior, although Ziff does claim that this is the way in which we discover about another
^5Paul Ziff, "About Behaviorism," The Philosophy of Mind, ed. by V.C. Chappell (Englewood Cliffs, New Jersey: Prentice- Hall, Inc., 1962), pp. 147-150.
56Ibid., p. 150. 79
person whether he is in fact gnashing his teeth or that
his teeth are simply gnashing. Ziff himself does not go
on to argue why such a claim that one finds out whether
oneself is gnashing his teeth or having his teeth gnash
by attending to his own subsequent behavior is odd. But
he does mention Wittgenstein in this regard and it is easy
to suppose that what Ziff has in mind here is that kind
of "oddness" ("meaninglessness") that Wittgenstein claimed many statements asserting knowledge about one's own mental
states had.5^ And, although Ziff makes no attempt to work
out the full-blown argument for the thesis that we know
our own actions in a way that other people do not, it has been worked out in detail by A.I. Melden elsewhere,5® and we shall presently be turning our attention to his argu ments .
Now, in some sense Ziff is right in claiming that the difference between someone's gnashing his teeth and some one's having his teeth gnash is to a large extent a matter of certain contextual and relational matters. The contex tual theory is obviously correct since everyone seems to agree that there is nothing in the movement itself which
^Compare "I know I'm in pain" v/ith "I know I'm six feet tall."
5®A.I. Melden, Free Action (London: Routeledge and Kegan Paul, 1965). marks the distinction between a human action and a (mere)
bodily movement. The major difficulty with Ziff's theory,
however, is his limiting of these contextual and relational
matters to the individual's subsequent behavior. In the
first place there is no reason why we should restrict our
selves merely to the agent's subsequent behavior. For as
we shall see in Chapter V, the agent's prior avowals and
planning or lack of such is certainly relevant to the
question of whether or not a person performed an inten- '
tional act or not, and it is certainly relevant also to the question of whether he can be said to have acted at all.
Thus, for instance, if a man is known to have uttered
numerous threats against another, his claim that he attacked him as the result of some muscle disorder is not likely to be taken very seriously, especially if he has no history of
such muscle seizures. Again, exactly what does Ziff mean by the- word “behavior"? Exactly what are we to look for in the way of subsequent (and we may add prior) behavior in order to settle the question of whether, on any particular occasion, a person is gnashing his teeth or merely having his teeth gnash? It is perhaps significant that nowhere does
Ziff tell us what this behavior must be. It is significant because it is unlikely that he could cite any specific sort of behavior which would settle the question of whether on a particular occasion a person was gnashing his teeth rather than having his teeth gnash. Finally, although everyone seems to admit that there is nothing in the move ment itself which marks the distinction between my gnash ing my teeth as opposed to merely having my teeth gnash, it does not follow that this distinction does not consist in there being relevantly different causes of the move ment involved. And this difference in causation need not be the difference between physical causation and mental causation (whatever that is). That is, Ziff has done nothing to show that the difference between a simple bodily action and a simply bodily movement does not consist in a difference in the way in which the movement is caused.
Thus, while Ziff's theory is illuminating in that it suggests that the difference between a bodily action and a
(mere) bodily movement is not to be discovered in the move ment itself but at least in some cases in certain contextual matters, his failure to specify these contextual matters in any clear fashion leaves the nature of a human action as mysterious as it was in the beginning. 82
2.9. A. I. Melden*s Theory
One of the most influential and at the same time il-
luminating attempts at elucidating these contextual ele ments which account for the difference between (mere) bodily movements and human action, can be found in the works of A. I. Melden. Perhaps the most celebrated of these works is Melden's little book Free Action. In this work, Melden attempts to defend the common-sensical belief in free human action against the doctrine of a complete causal determinism, and in doing so he succeeded, I believe, in combining some of the insights of Ascripticism and Con- textualism. That is, he saw clearly that a great deal of the difference between a (mere) bodily movement and a human action was somehow intimately tied up with the notion of ascribing responsibility to an individual.
The usual interpretation of Melden's positive con tribution to the elucidation of our everyday concept of a human action is that he claimed that a human action is a piece of "rule-following behavior." Thus, for instance,
Myles Brand has interpreted Meldenrs position as follows:
Human action can only be understood in the context of all human endeavors, or as Melden puts it in his paper "Action," ". . . i n the 83
context of practices in which rules are obeyed, criteria employed, (and) policies . . . observed.”59
And from this Brand surmises, “An action, then, is performed only if a rule is followed.”5^ Accepting this as the cor rect interpretation of Melden's view, Brand offers two ob jections to it. First, it is perfectly possible for a person to act without following a rule as is evident in the case of artists, social reformers, inventors, and innovators who deliberately act so that they are obeying no established practices, traditions, or rules. ^ And, secondly, granted that a knowledge of the relevant customs, practices, and rules that a person may be following in performing an action may help us to say what sort of action it is, “. . . it does not permit us to say what an action iis, which is the onto logical question we have been trying to answer.”52
But surely this interpretation of Melden and the criti cisms based upon it are mistaken. In the first place, to
59Myles Brand, The Nature of Human Action, op. cit., p. 17.
61 Ibid.
62Ibid. suppose, as Brand apparently does, that Melden believes that an action is performed only if a rule is followed is to greatly oversimplify what action theorists like
Melden, Benn, and Peters are trying to say. It is, of course, not difficult to thumb through the works of these philosophers and find places where they seem to be saying exactly this. Thus, Peters, for example, states in The
Concept of Motivation that "Man is a rule-following animal,11 and "Man in society is like a chess-player writ large."^
And Melden both in his earlier paper "Action" and in his book Free Action, explicitly compares the question "What is 6 t: a human action?" and the question "What is a chess piece?"
But a more careful reading of these works will indicate that neither Melden nor Peters suppose for a moment that this is the whole story. Hence, in The Concept of Motivation,
Peters remarks, "The paradigm case of a human action is when g g something is done in order to bring about an end,"00 adding
63 R.S. Peters, The Concept of Motivation, (2nd ed.; London: Routeledge and Kegan Paul, 1960), p. 5.
64Ibid., p. 71. 6 k A.I. Melden, "Action," in The Nature of Human Action, ed. by Myles Brand, op. cit., pp. 93-94. C f . Free Action, op. cit., p. 4. Italics mine.
66R.S. Peters, op. cit., p. 4. that "... norms enter into and often entirely define the en(j. "67 Likewise, Benn and Peters claim, "... a human action is typically something done in order to bring about a result or in accordance with a standard."*’8 A careful reading of Melden's works, I believe, will reveal much the same thing. Hence, although Melden does compare the ques tion, "What is a human action?" with the question, "What is a chess piece?" he is careful to point out that this is a 69 relatively artificial case. In dealing with possible con fusions which might arise with the analogy between a chess move and a human action, one of them being that we may speak of a person performing an action only if there is some rec ognized rule which he is following in doing so, Melden states,
All this may be granted, but it will be objected that a chess move is only one very special kind of action. We act in all sorts of ways, even in sweeping, the chessmen off the board, thus bringing the play to an abrupt end. With this I should
67 Ibid., p. 5. (Italics mine.) 68 S.I. Benn and R.S. Peters, Social Principles and the Democratic State, "Freedom and Responsibility" (London: George Allen and Unwin, Ltd., 1959). Reprinted in The Prob lems of Philosophy, ed. by William P. Alston and Richard B. Brandt (Boston: Allyn and Bacon, Inc., 1967), p. 352.
®^A.I. Melden, "Action," op. cit., pp. 93, 95-96. 86 certainly agree, but the case of the chess move is nonetheless important, for the very artificiality of the example may serve to remind us of what is too easily forgotten in the case of other types of action, namely the crucial importance of the prac tical context of common or shared practices involved in following rules, applying criteria, observing principles, acting on policies, and so on. Actions do constitute a whole family of cases, but in var ious respects this practical context is essential to an understanding of the distinction between a bodily movement and an action.70
Another crucial mistake made by people who interpret
Melden as asserting that there is an action only where there is the procedure of following a rule is that they neglect to notice that when Melden speaks about human actions he not only speaks of the existence of a rule or convention, he also makes a point of mentioning the circumstances in which an act is performed. It is seldom, if ever, that the exist ence of a rule alone together with a piece of behavior is sufficient for someone to have performed an action. But frequently it is a matter of there being an appropriate rule, an appropriate piece of behavior, and the right circumstances.
Thus, for example, S's extending his arm out of the window would not be a case of signalling if S was not driving his car at the time,if, say, the car was sitting in his closed garage at the time. Nor would S's moving his queen to king-
^ I b i d ., pp. 95-96. knight-seven constitute checkmating his opponent, if the
positions of the other pieces had been significantly different. Hence, it is not just the existence of an ap
propriate convention or rule which constitutes the only difference between a bodily movement and an action, but
the existence or nonexistence of certain circumstances is
also relevant. Indeed, not only is it the case the circum
stances are relevant to the question of whether a particular
sort of action is performed which depends for its intelligi bility on the existence of certain rules or conventions (e.g
signalling, checkmating), it also would appear that some actions require only a certain piece of bodily behavior plus the right sort of circumstances. If we suppose that some one performs an action only if there is a rule which applies to his bodily movement in these circumstances, then it would appear that taking a walk, washing the car, opening a door, eating an apple, and raising an arm are not instances of action. It would, I think, be quite difficult to discover rules for each one of these actions. What would be the relevant rule for taking a walk, washing the car, opening a door, eating an apple, or raising an arm? But, on the supposed analysis, no rule - no action. Notice, however, that although not all acts presuppose rules, many and per haps all acts require some set of appropriate circumstances.
It is difficult, for instance, to see how Jones' coming home at three in the morning could constitute breaking his 88
promise to his wife unless Jones promised his wife that
he would be home before three. And more importantly, an
agent's reasons (e.g., his intentions, desires, motives,
etc.) may be construed as the circumstances which determine
the action he performs. Hence, Smith's dangling a line in the water does not constitute fishing unless he does so because he wants to catch fish, and crouching behind a bookcase does not constitute hiding unless the agent intends thereby to escape someone's notice. Therefore, "taking a walk," "washing the car," "opening a door," "eating an apple," "raising an arm," "hiding," "fishing," "answering the telephone" are all ways of speaking which involve reference to certain sets of circumstances, and not, obviously, to any set of rules or conventions.
Melden rejects the attempts at analyzing human actions in terms of psychological factors, bodily movements, bodily movements plus some antecedent event (mental or physical), and behavior plus dispositions to behave. It might well be asked then what it is that he puts in their place. Yet such a question, Melden believes, is potentially misleading. For if what is meant by an analysis here is a series of state ments stating the necessary and suffucient conditions for the correct application of action terms, then Melden is quite 71 explicit in claiming that no such analysis can be given.
^A.I. Melden, Free Action, op. cit., pp. 193-195. 89
Indeed, so implausible is such a suggestion to Melden that 7 2 at one place he compares it to barking at the moon. * it is an essential part of Melden's view, then, that there simply isn't any answer to the question "What is a human action, simpliciter?" But this is not to say, says Melden, that the concept of a human action is a primitive, an indefinable something, and he sets himself the task of pro viding at least some sort of minimal elucidation of our everyday concept of a human action.
Melden's attempt at elucidating the contextual elements of our ordinary concept of a human action cannot be under stood unless one realizes that, for him, the concepts of a human action, having a reason for acting, and being a human agent, are all intimately tied up with one another.
According to Melden, reasons can be said to explain actions in a two-fold fashion: First, they explain by defining or identifying the action performed, and second, they explain by telling us something about the agent him self.^3 The first sense in which citing an agent's intention
72Ibid., p. 198.
73Ibid., p. 102. 90
or motive can be said to explain his actions can be easily
seen in terms of the following stock illustration. Con
sider someone who raises his arm as he drives his car toward
an intersection. Now, although it is not usually asked why
he raised his arm, it is conceivable that someone unfamiliar with the situation might ask. If we answer that he did so in order to indicate to others that he was preparing to make a turn, this answer enlightens us because it gives us a more precise notion of exactly what it is that the agent is doing.
Or, as Melden sometimes puts it, citing of the reason re describes the action of raising his arm as the action of signalling for a turn. It
. . . tells us what in fact the person was doing. It informs us, qua motive, that the action of raising the arm was in fact the action of giving information to others to the effect that the driver was preparing to make a turn.?4
This sense in which citing an agent's reasons explains his actions will be a feature of my consideration of the so- called "logical connection" argument, and I shall not stop here to discuss it further. It will be discussed in detail in Chapter III.
It is primarily the second way in which citing a reason can be said to explain an action which Meldon seems to want
74Ibid., p. 88. 91 to emphasize as being of central importance to understanding how reasons explain actions. That is, somehow the notion of acting and acting for a reason are intimately tied up with the notion of being a human agent. But just how does citing a reason reveal something about the agent himself and in so doing help to explain the action?
In attempting to answer this question, Melden points out that there are many different borderline cases between performing an action and a mere bodily movement (e.g., a man half asleep or coming out of narcosis, an infant who lacks in part the primitive ability to move its limbs at will, a madman, and so on) but that if we desire to under stand what is going on when an agent is acting from a reason, we will be forced to concentrate our attention on what hap pens when a fully conscious agent acts for a reason, i.e.,
"an agent who is intelligent in and attentive to what he is doing."75 what we see in such cases, Melden believes, is that our understanding of the concepts of a human action, having a reason for acting, and being an agent, are all intimately tied up with one another. The essential feature here seems to be the characteristic of consciousness or attending. Hence, Melden claims,
^5Ibid., p. 135. 92
. . . it is self-contradictory to suppose that in our full-bodied sense of these terms there could be agents who never act attentively and for a reason...only a person who has been attentive can be said to act for no reason and without attending to what he is doing."®
Rather, one must understand by this that there is a rock-bottom matter involved in the very concept of an agent - one who by using his limbs is able to do things in the presence of and with the things to which he attends. This is not a matter to be explained in terms of something else - rather it is our starting point in any explanation we give of conduct.77
For where there is wanting there is an agent who does; where there is doing there is an agent who can be and has been attentive to what he is doing and who acts as he does for a reason; where there is doing for a reason there it is false that one does what one does not want to do.7^
Actions, and the desires that explain them, are of course the actions and desires of agents; but this prepositional phrase does far more than indicate an identifiable subject or owner. Rather, it serves to mark the logically important point that the character of the concepts of action and desire and the kind of agent - man or animal - are reciprocally dependent upon one another. It is for this reason that the consideration "What shall we make of a man who stuffs his room with faggots, for no reason at all?" is crucially important. The fact that were this to occur we should be utterly bewildered shows that important features of our familiar concepts of both action and agent simply will not fit this case - we would not understand the action because we simply could not understand such an agent.79
76Ibi d., p. 137.
77Ibid., p. 139.
7®Ibid., p. 141.
79Ibid., p. 151. 93
Here the explanation of the man's conduct in terms of his desire turns out to be an explana tion of conduct in terms of his intention which gives us an account both of the man - his in terests and hence the good with which he is concerned - and his conduct.
A traffic policeman will perhaps consider per sons simply as motorists, passengers or pedes trians , as he discharges his duties. But our interest in knowing why the person in question is signalling is not so circumscribed. For the answer to this question which states the motorist's intention - let us say that he does this because he is on his way to the store on the street into which he is about to turn - enables us to understand what the motorist is doing by specifying his further status as a person: he is on his way to make a purchase.81
Now if I understand all of this-, it means that Melden believes that reasons can succeed in explaining actions because reasons are essentially tied up with agents. In other words, in order to come to a full understanding of what a human action is, it is necessary to understand what it means to say that something has purposes, desires, goals, intentions, etc., and in order for us to understand what it means to assert that something has purposes, desires, goals, intentions, etc., we must understand what it means to assert that something is a human agent, i.e., a being which is at least sometimes conscious and attentive to what he is doing.
80Ibid., pp. 154^155.
Slibid., p. 191. 94
But this answer would again appear to be subject to the
accusation of obscurity unless some kind of an analysis
of the concepts of persons and consciousness is forth
coming which shows that what people do is act rather than
merely behave. Hence, it would seem that this method of
elucidating the concept of a human action will depend for
its validity on a satisfactory solution to the mind-body
problem.
It is undoubtedly just this feature of consciousness
or awareness which accounts for our willingness to speak of
the movements of individuals as actions and our reluctance
to extend this talk to inanimate things. It is not unusual,
for example, to metaphorically describe the movements of a glacier as purposeful (e.g., as attempting to reach sea-
level) , but it is unclear how we might attribute desires, needs, aims, purposes, feelings, etc. to such things. Our
reluctance to ascribe desires, needs, aims, purposes,
feelings, etc. to such things springs from the fact that there does not seem to be a clearly understood sense in which we can speak of such things as being aware of their circumstances and what they themselves are doing. And at least one of the primary reasons why we find it so strange to ascribe awareness or consciousness, and consequently, desires, purposes, emotions, etc. to such inanimate things 95
is that they simply lack the necessary means of awareness.
They lack, for instance, the necessary eyes for seeing, ears
for hearing, noses for smelling, skin for touching and
feeling cold, heat, or pain, and so on.
Of course, it is possible to intelligibly ascribe such
predicates implying consciousness to inanimate things, but
such talk only becomes intelligible if we antecedently at
tribute some means of acquiring awareness to them. Such
metaphorical uses of conscious predicates then seem to be
parasitic on their uses to attribute consciousness to things which have the necessary means of being aware. Wittgenstein
seems to have held much the same view, except his account deals also with the characteristic of having the ability to
talk as well. In the Philosophical Investigations he gives the following account of such a case:
But doesn't what you say come to this: that there is no pain, for example, without pain- behavior? - It comes to this: only of a living human being and what resembles (behaves like) a living human being can one say: it has sensations; it sees; is blind; hears; is deaf; is conscious or unconscious. But in a fairy tale the pot too can see and hear? (Certainly; but it can also talk.) But the fairy tale only invents what is not the case: it does not talk nonsense. - It is not as simple as that. Is it false or nonsensical to say that a pot talks: Have we a clear picture of the cur- cumstances in which we should say of a pot that it talked? 96
Couldn't I imagine having frightful pains and turning to stone while they lasted? Well,how do. I know, if I shut my eyes, whether I have not turned to stone? And if that has happened, in what sense will the stone have the pains? In what sense will they be ascribable to the stone? . . . Look at a stone and imagine it having sensations. - One says to oneself: How could one so much as get the idea of ascribing a sensation to a thing? One might as well ascribe it to a numberI - And now look at a wriggling fly and at once these difficulties vanish and pain seems able to get a foothold here, where before everything was, so to speak, too smooth for it.82
Thus, the capacity for awareness as reflected in having the necessary equipment for awareness is an essential criterion for ascribing consciousness, including purposes, aims, goals, desires, feelings, etc. to individuals. And, if Melden is correct, it is also a necessary requirement for performing actions in the full-blooded sense and ascribing actions to others.
Yet merely possessing the necessary apparatus for aware ness cannot in itself be the whole story about why we are tempted to attribute purposes, aims, desires, feelings, actions, and so on to human beings while we feel no such temptation with regard to other things. For if the capacity and perhaps the actual use of these means of awareness were the sole determinates of this, then the philosophical problem
8 2 Ludwig Wittgenstein, Philosophical Investigations, trans. by G.E.M. Anscombe (New York: The Macmillan Company, 1958), Sections 282-284. 97
about whether sophisticated machines, which for all intents
and purposes do succeed in duplicating an.enormous variety of human performances, have thoughts, desires, aims, feelings, and perform actions rather than mere movements, would have been resolved long ago. For it is clear that such machines do have the necessary receptors of awareness. The sug gestion that there must exist some irreducible difference between such machines and men because men have the capacity to succeed in achieving a goal in completely novel circum stances and in insightive and original ways while machines are doomed to limited and inflexible patterns of behavior, is basically mistaken. For if I am not mistaken, machines have now been perfected which do in fact invent and instigate completely new strategies for dealing with complex problems? strategies never before thought of by human minds. Thus, for instance, complex computers have invented completely new and ingenious strategies for winning at chess. And even if this had not already been achieved, there is no good reason to suppose that machines could not be constructed which would be able to perform in the required fashion. Nor would it seem that the manner of manufacture is a relevant difference between men and machines. Being constructed out of synthetic materials does not automatically preclude membership in the family of man. Being "born of woman" seems to be nothing less than a mere prejudice similar to 98 the prejudices involved in having or failing to have a certain ancestral and/or racial origin. And seeing as it has been the case in some parts of the world and in some fields of endeavor that such ancestral and racial prejudices have been successfully overcome, there is no good reason to suppose that the same thing cannot occur with regard to 83 robots as well.
In addition to these rather commonplace objections against machines based upon quality of performance and means of manufacture, Richard Taylor has expressed some reservations about the practice of ascribing purposes, aims, and goals to machines in general. In his book Action and
Purpose, Taylor examines what he takes to be a perfect example of a "purposeful" mechanical object, i.e., a target- 84 seeking torpedo. In his treatment of such goal-seeking mechanisms, Taylor considers three different types of such torpedoes: one which is self-propelled and is guided by the sound waves of its target’s propellers; one which is self-propelled and guided by a cable attached to the target? and a third which is not self-propelled but is attached to the target by a rope so that it will drift into the target and explode when the target stops. According to Taylor, the
®^Cf. Hilary Putnam, "Robots: Machines or Artificially Created Life?" in Philosophy of Mind, ed. by Stuart Hampshire, (New York: Harper and Row, 1966} pp. 63-91. 84 Richard Taylor, Action and Purpose, op. cit., pp. 227- 233. 99
essential moral about all three of these types of goal-
seeking mechanisms is that in none of them is it literally
true that they are self-directed. The only sense in which
we may speak of them as goal-seeking mechanisms is a deri
vative and parasitic sense. That is, they can be spoken
of as goal-seeking mechanisms only in the sense that they
are designed and employed in order to satisfy certain
purposes and intentions of certain men. What this means is
that the torpedoes cannot be said to have purposes or goals
in themselves simply because they have no interests of their
own. From this Taylor concludes that although it is possible
to impute ends and goals to whatever displays the appropriate
type of behavior, it is always an illusion in regard to in
animate things, but only sometimes an illusion in the case of men.
Taylor's way of arguing here suggests that the essential feature behind the refusal to literally ascribe ends and goals to mechanisms is the lack of any coherent way of at tributing any independent interests to them. Yet I doubt that having independent interests can be an absolutely es sential requirement for acting with a purpose. For, in the first case, there is a certain ambiguity in the whole notion of dependence and independence. It is by no means self- evident that one man's interests are always or even most of the time totally and irrevocably his own. A young man's interest in becoming a doctor may be influenced by his
parents' aspirations and desires as well as by his friends
and relatives. And secondly, it is by no means clear that
we have any clear way of finding out in a particular case
whether the goals and ends that we think we see in some
thing's behavior is a result of its own interests or not.
Suppose, for example, that we were to discover on some
distant planet a race of machine-like creatures with no
discernible masters or controllers. No doubt, in this
case it would be quite natural for us to assume that even
though there are no discernible masters or designers
present now there must have been some at some previous time.
But supposing that these robots act in every way exactly
like us - they have established forms of government, economy,
and industry, they even bear and rear children in the normal
fashion, etc. - would we still be inclined to say that they
could not possibly be acting from interests and goals which were their own? The answer, I believe, is no. Of course, it may still be claimed that whatever interests and goals they have are in some sense derived from their designers1 interests and goals, but there hardly seems to be any com pulsion to do so. 101
Finally, just as it is extremely difficult to formulate defining criteria for something's being a man, it seems no less difficult to formulate non-question begging criteria for what is going to count as a machine. The materials of construction may sometimes serve as a means of determining whether something is a machine or not, but it is doubtful that it can always serve as a defining criterion in itself.
This becomes extremely apparent if the machine just happens to be constructed out of protein. The method of construction itself may also serve as a defining criterion. But as I have already argued, being born in the natural way is, at most, a mere question-begging prejudice.
It would seem then that if the supposed distinction between actions and mere movements depends upon the presence or absence of consciousness or awareness, there is no clear line which can be theoretically drawn between what we would consider a creature capable of thoughts, desires, goals, interests and feelings and those which are not.
In attempting to elucidate our concepts of human action,
Melden puts a great deal of emphasis on the training we receive as children. That is, he believes that the best way to understand our everyday action concepts is to realize that they are acquired largely through training, including the training that we receive in the use of the language 102 85 itself. Children typically receive training in responding to others in familiar transactions, and through this training they come to recognize others as persons and their bodily movements as actions. It is important to realize here, says
Melden, that in going through this process of training the child does not acquire a concept of an agent or an action simpliciter, but instead acquires the concept of specific agents and specific actions. Thus, as Melden argues,the child’s concept of a mother or a father, for instance, is hardly the same as the physiologist's or geneticist's. The child's concept of his mother is not a concept of some creature who has such-and-such a bone structure and such- and-such glands arranged in such-and-such a fashion, nor is it a concept of a creature with such-and-such genes and such- and-such chromosomes. Nor is the child's concept of his mother the concept of a female parent.
Innumerable skills and abilities are acquired by the child in the course of this training, and in order to pick up these various skills and abilities the child must have the ability to
. . . recognize that it is being commanded, comforted, fed, asked, played with, taught,
85A.I. Melden, Free Action, op. cit., pp. 193-197.
86Ibid., pp. 196-197. 103
succoured, questioned, clothed, etc., (and its role in these matters), by those whose status within the family are defined by the manner in which they carry on these proceedings with it. 7
It is just this training which is received by the child in his day-to-day transactions with others which determines the particular meanings which he attaches to the concepts of specific persons and their actions. To suppose, claims
Melden, that somehow there lurks some common notion of a person or of a human action (behind these concepts of specific persons and specific actions) is to ignore that when we come down to the training we have received in our early childhood, we have come down to rock-bottom. It is perhaps too easy for a philosopher to imagine that when he is attempting to analyze our ordinary concept of a person and a human action he is really attempting to discover some thing new about the nature of the world. But the truth is that beyond pointing to the training that all of us have received as children there is nothing left to be pointed to as a further justification of why we use these concepts as we do.
Meldenfs claims about how we come to acquire the con cepts of a human person and a human action are, however, insufficient to show that any unbridgable gap exists between
®7Ibid., p. 196. 104 ordinary language and the language of physiology. For if the intent of Melden's argument is that our everyday concept of a person and a human action are learned through the use of some specific paradigms, then Melden has failed to show that this process of learning the natural language is, in any significant sense, different from the way in which any other language is learned. That is, Melden's arguments are not sufficient to show that the language of psychology or the language of physiology are not learned in terms of certain paradigms. For, after all, it may be argued that terms like "stimulus,'’ "drive reduction," "thing," "process,"
"movement," "cause," and so on, are learned in terms of cer tain standard or peradigm instances. Hence, in pointing to the type of "training" which is used in teaching the concepts of action and person, Melden has failed to show that there is any radical incompatability between the concepts of phys iology or psychology and the concepts of ordinary language.
Melden, in Free Action, argues that our ordinary concept Op of desire or want borrows its meaning from the desirable.
If one could conceive of a pure intellect, Melden says,
. . . which lacked the experience of one who sees something called "food" as something to be eaten, which was incapable of thinking about
®®A.I. Melden, Free- Action, op. cit., pp. 160-170. 105
that kind of object as something to be relished or avoided, to be regarded with the feelings - the tensions, urges, impulses, etc., of one who does want food and gets or decides for one rea son or another not to get it - such a being simply would not understand "There is a rest aurant nearby where food is obtainable" in the way in which we do when we recognize that it is a reason for doing.88
And A.R. Louch has argued in much the same fashion that
understanding how reasons explain action is to understand how statements of reasons justify or warrant certain actions?8
Consider, for example, the way in which citing the agent's
intention in acting may succeed in explaining his actions.
One of the features which Louch mentions as relevant to our ascribing intentions to ourselves and others is that we see the action "in the light of a certain situation."
Seeing it in this light, Louch tells us, is "seeing it in the light, also, of some vaguely defined collection of human practices, institutions and rules." But just how do rules, institutions, and practices get involved in assigning inten tions? So far as I can determine, Louch brings in these features in the process of trying to explain how intentions
89A.I. Melden, Free Action,' op. cit., p. 166. (Italics mine.)
88A.R. Louch, Explanation and Human Action, op. cit. Louch does not mean by "Warrant," moral approval. See p. 52. (Berkeley and Los Angeles, California: University of Cali fornia Press, 1966), p. 109. 106 actually do explain actions. Thus, he claims,
. . . we need to explicate the sense in which an intentional description makes movements intelligible, and thus.explains them. The clear sense is the one in which an action is puzzling. In a very funny scene from a recent film a bicycle- riding postman is plagued by a bee. He is watched from a neighboring hillside by a puzzled farmer, who only sees the postman's frantic gestures. In a minute, the farmer discovers for himself why the postman indulges in his antics, for the bee comes up the hillside. It is as if we might say, when we observe the occasion of the action it is no longer puzzling. Seeing the bee entitles us to interpret the arm-flailing as "trying to brush off the bee." Hearing the doorbell ring entitles us to call the action "answering the doorbell." The relation of bee and doorbell to action is in such cases not causal; that is, our understanding the movements as intended is not merely another way of saying that the bee or doorbell caused the action . . . rather, the bee and doorbell provide the appropriate circumstance to such actions. We say, that is the sort of thing to do in such circumstances. The bee and doorbell, looked at this way, are logical cues to the action, for we think of bees as the sort of companions to be rid of, and ringing doorbells as the sort of signals for answering doors . . . The doorbell case leads directly to a convention. That is what bells are put there for. it might appear, on the other hand, that the buzzing bee does not mean flailing arms, but rather that the bee's sting creates a desire to be rid of the pesky insect. But even here, though we have taken the issue back to such rudimentary human experiences as pain and fear, the force of our account is the force of justi fication. The relation of pain and fear is not causal, for it does not involve research into regu larities , but depends upon identifying pain as that to be avoided.if possible. For all practical pur poses - except with untutored children, psychologists and sociologists - we might as well have stopped with the bee as the thing to be avoided . . . The connexion between the hurting and the avoidance is not like that between the sting observed in the abdomen and the pain and swelling its plunging into the skin causes. "Pain" involves "avoidance." 107
. . . So "it's pleasant" includes the notion of v/orthy-of-being pursued, "I want it" includes actions taken toward achieving (or trying), and "That's a bee," the notion of getting rid of it. . . We might then explain a person1s action by saying "the doorbell's ringing," or "A bee is buzzing 'round his head," without research or the unpacking of argument. Here it is pretty clear that the relation between the "cause" and the action is one of entitlement. The action is appropriate to the circumstances in accordance with either cases or criteria. Actions are movements seen and iden tified as warranted or not by circumstances . . .91
Now, if I understnad this correctly, the major point that Louch is trying to make is that to answer the question
"What is he doing?" by saying, for example, "He is swatting at a bee" or "He is answering the doorbell" is to describe what was done as an intentional act. And in doing this we are also partially explaining (justifying) what was done.
This is why saying "A bee is buzzing 'round his head," or
"The doorbell is ringing" explains what he did. But coming to see this does not require any special techniques of experimentation and generalization.
But exactly how do certain human conventions, institu tions, and practices come into the picture here? As Louch himself notes, they come into the picture first of all in quite a trivial way, since "the doorbell's ringing" and "a bee is buzzing 'round his head" are clearly ways of talking
91 Ibid., pp. 106-109. 108
which depend for their meanings on the existence of certain
conventions. But this is trivial, just because every piece
of language depends in some sense on the existence of certain
conventions for its meaning. But supposedly there is another way in which conventions, institutions, and practices enter
the picture which is not so trivial. The doorbell case is pretty straightforward. Doorbells are installed for the express purpose of signalling for the attention of the dwellers, and everyone understands their purpose. A ringing doorbell is a signal. But the bee case is a great deal more difficult to understand in this regard. As I understand it, a buzzing bee provides a reason for (explanation for, justification for) flailing arms just because it is a pretty universal human practice to avoid bees when possible.
To understand this in its entirety it is necessary,
Louch claims, to understand that attributing intentions to agents usually involves citing a desire. Hence, Louch claims,
But if we ask, what do we recognize, observe, or take note of what makes for intending or ex pressing intentions, we should have to appeal to a desire for something plus a knowledge as to what will achieve that end. So announcements of inten tion assimilate to the justifying procedures char acteristic of our explanatory use of desire. We assess situations in the light of the indefinite range of desirable objects and activities, and are thus led toqdeclarations of intention and descriptions of actions.
92Ibid., p. 113. 109
The claim then is that there are a whole family of desire concepts (i.e., liking, desire, need, pleasure, pain, and emotion), each of of which has some commonly accepted eval uative base.
Consider, for example, the concept of desire. The first thing to be noted is that there exists a certain basic ambiguity in the notion of desire. This ambiguity is best brought out in terms of an example. A woman shopping, seeing a hat in a store window, is overwhelmed by a desire to have it. She hesitates, thinks about the lecture her husband gave her about spending beyond their means, and finally, unable to resist it any longer, rushes into the store and buys it.
Nov/, Louch admits that there is a sense in which the hat can be said to cause the desire, and the desire can be said to cause to action. This is the sense in which it makes sense to say that if she had not seen the hat, she would not have been tempted, and if she had not been tempted, she would not have bought the hat. Therefore, Louch admits that a causal account or desire may be possible. But he insists that it does not follow from this that desire is thereby an object of scientific inquiry. It does not follow, he believes, simply because it does not require any special techniques of experimentation and generalization to discover when it is true to say that someone desires something. 110
Yet, there are still difficulties involved in recog nizing and identifying desires. That is, it is not merely a matter of identifying some object (the hat) or some bodily feeling or sensation, but it involves recognizing that the object or bodily sensation has a certain relationship to the agent. Louch points out three possibilities here:
(1) If the object is embraced or consumed, or used in some way, it will dissipate pain or tension (need).
(2) If the object is used in some way, it will prolong or intensify certain feelings.
(3) The object is judged to be worthy of pursuit on the basis of precisely worded criteria (e.g., wine-tasting).93
But in spite of the differences between these three cases, they may be seen as dependent on the agent1s actually recognizing the object desired as desirable, as worthy of being desired. To explain an action then, by appealing to the agent's desire, is to give what amounts to an evaluative explanation. That is, appeals to desires and wants are justifying explanations. The same, it would appear, applies to other reasons as well.
Such claims would seem to create an insurmountable impasse between those theorists who claim that a totally adequate nomological explanation can be given for human
^ I b i d . , pp. 66-67. Ill actions and those theorists whof because of this evaluative element, deny it. And thus the supposition has arisen that there are two incompatible conceptual schemes: our ordinary conceptual scheme with its concepts of reasons, actions, and persons, and the scientific conceptual scheme with its con cepts of behavior, stimulus-response, causes, brain events, and so on. But this impasse can be seen to be only an ap parent impasse once it is realized that what counts as an explanation is determined in part by pragmatic consider ations. Many action theorists seem to be opposed to causal accounts of human actions on the grounds that what results from such accounts is only the explanation of the bodily movements of pseudo agents, rather than the full-blooded actions of conscious agents. Thus, they claim that an ex planation in terms of events and empirical laws relating events explains why there occurred a forward movement of his left leg followed shortly thereafter by a forward movement of his right leg, but they insist that this is no such explan ation of the agent's actions, e.g., he walked to the grocery store, he signalled for a turn, or he recited The Raven.
If these claims are taken seriously, it would indeed, appear that there is an irreducible gap between our common- sensical explanations of human actions and scientific explan ations. For all explanation, whether it is scientific explan ation or the common-sensical type that we ordinarily offer of 112 an agent's actions in terms of reasons, clearly depends upon certain procedural rules which serve to define what will count as an appropriate explanation. Thus, it would appear that within our common-sensical framework, citing an agent's reason for acting constitutes an explanation of what he is doing because we operate on the procedural rule that, barring any intervening factors, a human agent will do what he in tends to do. While, on the other hand, the dominant proced ural rule within the scientific framework is to seek out true empirical generalizations which have reliable testable results.
There can be little doubt that such procedural rules themselves embody certain evaluative elements, and greatly influence how the phenomenon in question x^ill be described in the first place. How an individual phenomenon is described is a function of what the person doing the de scribing wants to do with it. Thus, if the person is inter ested in finding some empirical correlations between what he is observing and other events which he can test for, he will describe the phenomenon in terms which will provide the most favorable prospect of success. If, on the other hand, he is not interested in finding any empirical correlations be tween the phenomenon in question and other empirically test able results, then he is likely to choose some other way of 113
describing the phenomenon. Looked at in this way, it is
easy to see why so many action theorists have insisted that
causal accounts only explain bodily movements and not actions.
But even if such pragmatic elements do influence what
is going to count as a correct description and explanation
of something, it does not follow that no completely causal
account of human actions is possible, nor that our common-
sensical explanations in terms of reasons enjoy any great advantage over a more scientific explanation of them. For, if it is claimed that most, if not all, of the current ex planations of human behavior to be found in the social sciences are at most only explanations of bodily movements, then this situation can be remedied by providing transforma tion rules for translating talk about bodily movements into talk about actions. X believe that such rules are available and they will constitute part of the subject matter of
Chapter IV.
Furthermore, it is clear that both Melden's emphasis on early training and Louch's attempts at explaining how reasons justify or warrant certain actions only make sense against a general background of empirical facts about human beings. A buzzing bee only provides a reason for (explanation for, justification for) flailing arms because we are already aware of the general truth that people usually avoid potential sources of pain and suffering. Given this, it is no more
mysterious why human beings tend to avoid certain situations
and engage in certain activities than it is that two op
posite charged particles tend to repel each other. Nor is
it obvious that reasons considered in this sense are any
more evaluative than the conditions which are responsible
for two particles repelling one another. Perhaps there are
some actions which depend for their intelligibility on the
existence of certain social rules and conventions such as
issuing a check, but it is nevertheless true that we can
understand a person's reluctance to write a blank check
because we can see that such a course of action could be
detrimental to some pretty basic and common human desires.
Thus, rather than replacing empirical generalizations about
human nature, Melden's and Louch's analyses of reasons for
acting presuppose them.
And finally, it should be pointed out that Louch's claim
that desire is not an object of scientific inquiry because it
does not require any special techniques of experimentation
and generalization to discover when it is true to say that
someone desires something is simply false. For, in the first place, not all scientific inquiry or discovery of causes in volves the use of special techniques. The discovery that decapitation usually results in death did not require any
special technique of experimentation to discover. Secondly, 115 as I have argued earlier, it is by no means clear that there is not at least an unconscious inference involved in our coming to know our reasons for acting. Likewise, it was pointed out that reasons only seem to make actions intel ligible because they already presuppose a background know ledge of some pretty basic empirical generalizations about human nature. And third, special techniques are frequently employed by psychologists and psychiatrists to discover a patient’s unconscious desires where these desires cannot simply be readily "read off" from the patient's actions.
In this chapter I have examined a number of attempts at analyzing or elucidating the concept of a human action and have found them wanting in one way or another. It would be rash to claim that the elimination of these candidates in itself shows that no completely adequate analysis is possible. Nevertheless, their elimination does suggest that at least for the present there is no one all-encompassing analysis available. I have already indicated that I do not believe that such an all-encompassing analysis is possible, although I do believe that the notion of a basic bodily action can be adequately defined. This task will be per formed in Chapter V after the necessary groundwork has been completed. CHAPTER III
EXPLANATIONS, REASONS, AND CAUSES
In Chapter II, I examined a number of attempts at arriving at a satisfactory distinction between (mere) bodily movements and human actions. First, it was noted that there seems to be no one common intuitive notion of what a human action is. Next, Richard Taylor's arguments for the thesis that the notion of a human action was an unanalyzable ultimate notion were examined and found to be faulty. After this, a naive form of behaviorism whereby a human action is viewed as identical with an appropriate piece of behavior and/or the appropriate effect of this behavior was seen to be subject to a number of serious counterexamples.
The volitional theory was then examined and while it was argued that none of the stock criticisms of the theory are in themselves conclusive, there appears to be no independent evidence for the existence of such things as volitions.
Hart's attempt at providing an ascriptive theory of action was also examined and found to be inadequate, essentially because the ascriptive-descriptive distinction is not as clear-cut as has been imagined and the notion of responsi bility is too narrow to cover all of the required cases.
116 117
Likewise, it was pointed out that the notion of an "ascrip
tion of responsibility" is extremely obscure, . and that if one construes the notion of an "ascription of responsibility" in the broad sense in which Joel Feinberg does, the theory does not rule out causal or dispositional accounts of human actions. Next, Paul Ziff's attempt to account for the dif
ference between (mere) bodily movements and bodily actions was considered, and although it was argued that in some sense contextualism must be correct since everyone agrees that there is nothing in 'the bodily movement itself which distinguishes mere bodily movements from actions, Ziff’s em phasis on the subsequent behavior of the agent is too vague and limiting to be of much real help in making the required distinction. Following this, some potentially misleading and definitely oversimplified interpretations of A. I. Mel den's labors at further elucidating the contextual elements which determine the distinction between human actions and mere bodily movements were exposed. And finally, Melden's attempt at elucidating these so-called contextual elements surrounding the notion of a human action were examined and found to be essentially muddled and inadequate. In this chapter, I will examine the supposed contrast between ra tional explanations and scientific explanations of human ac tions along with the two major arguments which have been offered to show that reasons cannot be causes. 118 3.1. Rational Explanation and Scientific Explanation
The question of whether reasons are causes or not is
frequently discussed in the context of the much broader question of what constitutes a proper explanation of human
action. Some theorists have claimed that the sense in which reasons can be said to explain actions is not the sense in which causes explain effects or laws explain events, while others have held to the opposite view. Now what I wish to argue is that the sense in which reasons are claimed to explain actions is extremely muddled and does not rule out the possibility of giving nomological explanations, causal or otherwise, of human actions.
There are, of course, many different kinds of explan ations (e.g., explaining how, explaining why, explaining the location, explaining how far, explaining how long, etc.).
But generally, anything is a possible candidate for being an explanation of something provided only that it is a possible answer to the question "Why?" Yet it does not come as a complete shock to anyone to discover that philosophers have for a long time been aware of the fact that "Why" questions are extremely ambiguous. Aristotle’s formulation of at least four different senses of "cause" may be taken as a case in point. It would appear that, with regard to human actions, there are at least two major senses in which "Why" questions about them can be interpreted. First, they may be interpreted 119 as asking for the motive or reason that the agent had in doing what he did do. This is, of course, the sense in which such "Why" questions are most frequently interpreted. The second sense in which such "Why" questions may be interpreted is the sense in which it is understood as asking for the general laws and initial conditions which together are suf ficient to yield, produce, or cause the resulting action.
Problems connected with explanation in general are of course by no means new, but during the last twenty years or so philosophy has witnessed a revived interest in these problems, especially as regards human actions, and many analytic philosophers have since exerted a great deal of effort in attempting to formulate precise statements about exactly what constitutes a proper explanation of human actions. Basically, the problem has brought about a divi sion of philosophers into two major camps, those who claim that all legitimate explanation must proceed in terms of general laws, and those who deny that general laws are neces sary for legitimate explanations. Hence, some philosophers
(A.R. Louch, W.H. Dray, R. Taylor, et al.) have argued that the everyday explanations that we offer for people's actions, and that the historian frequently offers for the actions of some historical figures, are completely satisfactory as they stand and do not require the addition of some supplementary law or set of laws in order to make them into satisfactory 120
explanations. Still others (C. Hempel, A.J. Ayer, J. Fodor,
E. Wagel, et al.) have argued that nothing short of an ex
planation in terms of some general law or set of such laws
is to be considered as a respectable explanation. William
Dray first coined the label "The Covering-Law Model of
Explanation" to refer to this view, and in what follows I
shall simply refer to this view as the "covering-law model."
The covering-law model of explanation is so well-known
in philosophical circles that it hardly warrants a lengthy
exposition. Included in the list of its most celebrated
proponents are C.G. Hempel, E. Nagel, and J. Kim. Roughly,
the proponents of the covering-law model claim that the
formal logical structure of all legitimate explanation is
as follows: The statement which correctly describes the phenomenon to be explained must be capable of being shown
to be the logical consequence of a set of true statements describing the specific relevant circumstances present at the time the phenomenon in question occurred and/or directly preceding the time at which the phenomenon occurred? plus a set of true general law statements relating these antecedent or contemporary circumstances to the phenomenon being ex plained. Or, as Hempel puts it in "The Function of General
Laws in History,the logical structure of the explanation
94 Carl G. Hempel, in Readings in Philosophical Analysis, ed. by Herbert Feigl and Wilfrid Sellars (New York: Appleton- Century-Crofts, Inc., 1949), pp. 459-471. 121 of an event of kind E is
(1) a set of statements asserting the occurrence of certain events C-L,...Cn at certain times and places,
(2) a set of universal hypotheses, such that
(a) the statements of both groups are reasonably v/ell confirmed by em pirical evidence,
(b) from the two groups of statements the sentence asserting the occur rence of event E can be logically deduced.95
Anything which can be shown not to have this logical structure, or, at least, not to be a sketch of it, is not to be considered a satisfactory explanation of the phenomenon in question.
However, it soon became apparent even to those who championed this model of explanation that this ideal structure was unduly restrictive even for stirctly scientific explan ations. For, frequently explanations are given in science not in terms of universal general laws of the form, "All A's and
B ’s," but in terms of statistical general laws. To remedy
Ibid, p. 460. (2) (a) was changed shortly thereafter to avoid undue embarrassment over the possibility of accepting an explanation at one time and then having to reject it later because the explanation turned out to be false. To avoid this difficulty, the explanans must be true rather than just well confirmed. Cf. Carl G. Hempel and Paul Oppenheim, "Studies in the Logic of Explanation," in Philosophy of Science, XV (1948) p. 137. 122
this situation many covering-law theorists like Hempel have weakened the formal requirements on explanations allowing that statistical laws may be substituted in place of the universal general laws. This results then in admitting a second form of acceptable explanation with the following logical structure:
(1) The Explanans:
(a) Ps (GfF)=q
(2) The Explanandura: Gi Provided that q is close to 1, the explanans are not to be said to logically entail the explanandum, but rather, "to confer high inductive probability on" the explanandum. 96 The introduction of this second form of explanation does not affect the covering-law model, however, since in both cases a law statement or series of law statements is required for the explanation regardless of whether this law statement or series of law statements is universal or statistical. But as against this theory of what counts as an ex planation William Dray has persuasively argued that there ^Carl G. Hempel, "Reasons and Covering Laws in Histor ical Explanation," in Philosophy and History, ed. by Sidney Hook (New York: New York University Press, 1963), p. 145. Cf. Carl G. Hempel, "Explanation in Science and in History," in An Introduction to Philosophical Inquiry, ed. by Joseph Margolis (New York: Alfred A. Knopf, 1968), p. 254. 123 is still another kind of explanation which is at the same time both a legitimate form of explanation and neither de pendent on nor reducible to the covering-law model. If I understand him correctly, Dray contends that at least some times historians can and do succeed in explaining the actions of historical agents by reconstructing the agent's motives and beliefs (both about the relevant features of the situa tion and about the means to be adopted in order to reach the desired end). This type of explanation Dray calls "rational explanation," and claims that it provides us with an under standing of the action by showing us that it was the reasona ble thing for the agent to do, as seen from the agent's own point of view. That is, what such explanations aim at is showing that the action which the agent performed was, from the point of view of the agent himself, i.e., given his particular motives, beliefs, and principles at the time, the "appropriate" thing for the agent to do. Dray's stock ex ample of such a rational explanation is Trevelyan’s explana tion of why King Louis XIV withdrew military pressure from Holland in the summer of 1688 even though the action did not in fact have favorable consequences for Louis.^ 97 Although Dray first presented and defined his notion of rational explanation in his monograph Laws and Explanation in History (London: Oxford University Press, 1957), I shall be concerned here only with his more recent presentation and defense in "The Historical Explanation of Action Reconsidered," in Philosophy and History, ed. by Sidney Hook, (New York: New York University Press, 1963), pp. 105-135. 124 But the truly significant thing about rational explana tion , Dray insists, is that the inclusion of suitable em pirical laws is neither a necessary nor a sufficient con dition for its being a legitimate explanation of the said action. It is not a necessary condition because the aim of the explanation is merely to show that the action did make good sense from the agent's own point of view, and it is not aimed at showing that the agent was the sort of man who always does this sort of thing when he is in the sort of circumstances which he thought himself to be in in this particular instance. The inclusion of some law in the ex planans is not a sufficient condition either since the deduction of the explanandum from the explanans will not show that the thing that the agent did was the reasonable or appropriate thing for the agent to do, given his parti cular purposes, beliefs, and principles at that time.^8 It is no part of my intention to attempt a complete critical examination of all of the particular points of disagreement between Dray and Hempel on this question. Many of these points have already been quite ably discussed by Dray and Hempel themselves. Instead, I shall concentrate my attention on what I take to be the most serious points ^William Dray, "The Historical Explanation of Action Reconsidered," in Philosophy and History, ed. by Sidney Hook, o p . cit., p. lo9. : 125 of contention between these two models of explanation. To my mind, perhaps the most serious point of contention be tween Dray and Hempel here is the issue of whether and to what extent the rational model of explanation involves or presupposes general empirical laws. 3.2. Belief and Reason Explanation It is perhaps too obvious to mention that anyone who requires that an adequate explanation provide good (induc tive or deductive)* grounds for asserting that the phenomenon in question did in fact occur will claim that neither rea sons nor beliefs alone are sufficient in themselves to ex plain why an agent performed a given action. Reasons (e.g., motives, wants, desires, aims, purposes, etc.) and beliefs seem to them to be both necessary for even a minimally ade quate explanation of human actions. We cannot, for example, explain why Clifford goes to the refrigerator unless we state both a relevant reason and belief. We cannot explain why he went to the refrigerator by saying simply that he wanted an ice cream bar; we must, in order to have a minimal explanation, also cite some relevant belief or other, such as he also believed that there might be an ice cream bar there. Likewise, by varying the beliefs and reasons we get a quite different explanation of why an agent performed the particular action that he did. Hence, we might explain why Jerome drank a glass of acid by saying that he wanted to 126 drink a glass of water and believed that what he had in his hand was a glass of water rather than a glass of acid. Or, we may explain it by saying that he wanted to collect his accident insurance and believed that he had a glass of acid in his hand and that drinking it would be sufficient to cause the required injury. There is perhaps a great deal of justification in the accusation that many action theorists are guilty of ignoring the essential role that an agent's beliefs play in explain ing his actions. Thus, not only is the role played by the agent's beliefs in the explanation of his actions very in frequently mentioned by action theorists in general, when they are explicitly mentioned they almost invariably im mediately drop out of the picture. For example, Melden in discussing a driver signalling for an approaching turn lists some of the things that he might think and feel as he raises his arm to signal. These might include such things as being aware of the fact that he is driving on the road, seeing an approaching turn, noticing that a car behind is following close, seeing a garage on the corner, feeling relief at the prospect of this being the last turn before home, feeling irritation over the delay caused by another turn, feeling the relaxing of the leg muscles as the foot 127 on is moved from the accelerator. J Melden's major point is that none of these things can be identified with the intention to signal, but regardless of how it is with regard to this point, it is surely the case that at least some, if not all, of the items on this list presuppose certain relevant beliefs on the part of the agent. One cannot, for instance, be aware of the fact that he is driving on the road without in some sense believing that there is a road, the car is on the road, he is at the controls, it is moving on the road, and so on. There is no necessity here to suppose that the driver, especially if he is not a novice at driving, will in any sense be consciously entertaining these beliefs in the form of occurrent thoughts at this time. No matter how beneficial it might be if some drivers occasionally reminded themselves that they are driving on the road, even a careful driver does not usually entertain the thought that he is indeed situated in the driver's seat driving down the road. The truly surprising thing about Melden's subsequent handling of this case is not that he denies that certain beliefs are presupposed in the situation, but that they immediately drop out of the picture. Instead Melden concentrates his attention on the reason or intention the driver had in raising his arm. " a .I. Melden, Free Action, op. cit., pp. 96-97. 128 1 As far as I can determine there are three possible ways of accounting for this apparent lack of emphasis on the role that an agent's beliefs play in explaining his actions. First, Melden may be simply rejecting the notion that a legitimate explanation must provide an argument asserting that the agent actually performed the action in question. Second, the implied beliefs are simply too well- known to deserve explicit mention. And finally, somehow we are informed of an agent1s relevant beliefs by being informed of his relevant reasons for acting. The first way of accounting for the apparent lack of interest in the role that an agent's relevant beliefs play in explaining his actions would, it would seem, simply bring us to a stalemate again between the covering-law theorist and the action theorist, each claiming that only their models succeed in "explaining” human actions. The second way of accounting for the lack of emphasis placed on beliefs is, I believe, quite revealing since, if I am not mistaken, many action theorists are led from the fact that great masses of individuals frequently share a large repertoire of beliefs in common, thus making it unnecessary and even pedantic to explicitly mention them in explaining an agent1s actions, to the notion that somehow reason statements, by virtue of their meaning; already contain a reference to the relevant beliefs of the agent. But besides 129 this being a fallacious move in itself, it seriously impairs the scope of reason explanations insofar as it simply is not true that all individuals nor all groups of individuals share exactly the same beliefs in common. So, contrary to the impression given by Melden's account, the role that varying beliefs play in the explanation and determining of action cannot be overemphasized. It is not totally implausible, for instance, that the behavior of certain neurotics and psychotics can be explained not only on the assumption that they do not share the normal desires and wants, but they may also be explained on the assumption that they also fail to share the normal beliefs. The man who takes elaborate and insane precautions to protect his life believing that everyone is out to get him would be a typical example. As psychologists and psychiatrists we may be interested in discovering how he came to hold such a mistaken belief, but it can hardly be denied that his belief is relevant to an explanation of his behavior. Likewise, although it is not considered to be insane behavior, the practice of leaving the old members of a tribe to die in the elements is made a great deal more palatable to our moral sense when it is discovered that the tribe universally believes that the victims will enjoy a pleasant and happy afterlife. The extremely important part played by an agent’s relevant beliefs cannot be ignored. 130 3.3. Dray’s Principles of Action Dray is one of those action theorists who recognizes that both reasons and beliefs are necessary for an explana tion of human actions. The real issue between Dray and Hempel is whether citing an agent’s relevant reasons and beliefs is in itself sufficient to explain the agent's action, or some relevant empirical statement is also required. Now it is clear that not even Dray himself believes that simply citing an agent's relevant beliefs and reasons is, by itself, sufficient to explain why the agent performed the action that he did. But what is needed here to complete the explanation is not, according to Dray, any true empirical generalization, but what he calls "principles of behavior" or "principles of action."100 Nevertheless, Dray's principles of behavior or action do have an element of generality in them, for as he explains, For it is quite true that "reasons for acting" as well as "conditions for predicting" have a kind of generality or universality. If y is a good reason for A to do x, then y would be a good rea son for anyone sufficiently like A to do x under sufficiently similar circumstances.101 But in spite of this element of generality or universality ^William Dray, Laws and Explanation in History, Chap ter V, Section 4. Reprinted in Readings in the Theory of Action, ed. by N.S. Care and C. Landesman, op. cit., pp. 268- 274. 101Ibid., p. 269. 131 that principles of action have. Dray insists that it is significantly different from the type of generality or universality that general empirical principles have. Hence, he argues, But this universality of reasons is unlike the generality of an empirically validated law in a way which makes it especially hazardous to say that by giving a rational explanation, an his torian commits himself to the truth of a cor responding law. For if a negative instance is found for a general empirical law, the law it self must be modified or rejected, since it states that people do behave in a certain way under certain circumstances. But if a negative instance is found for the sort of general state ment which might be extracted out of a rational explanation, the latter would not necessarily be falsified. For that statement would express a judgment of the form: "When in a situation of type Cj . . . cn the thing to do is x." The "implicit law" m such explanation is better called a principle of action than a generali zation (or even a principle of inference),1°2 According to Dray, the connection between these so-called principles of action and the cases falling under them is neither deductive nor inductive, but "intentionally and peculiarly loose." •*-02Ibid., p. 269. Apparently the parenthetical remark here is designed to dismiss the Instrumentalist's claim that some general principle is necessary for a satisfactory ex planation except that the principle itself is not to be con strued as an explicit premise in the explanation or pre diction, but instead as a rule of inference allowing us to move from the statement of the conditions directly to the statement of the occurrence of the phenomenon. I purposely avoid the criticism that because of this feature of princi ples of action, Dray's principles may be cognitively meaning less because they may be in principle unverifiable. 132 If Dray is correct here, it must be possible to show that the judgment, "What in a situation of type C]_ . . . Cn the thing to do is x" is not capable of being construed as an empirical generalization. And if I understand him correctly, the reason why Dray thinks that this cannot be construed as an empirical generalization is that the consequential phrase "the thing to do is x ” is primarily a value phrase. Thus, he says, The phrase "thing to have done" betrays a crucially important feature of explanations in terms of agent calculations - a feature quite different from any we have noticed so far. For the infinitive "to do" here functions as a value term. I wish to claim therefore that there is an element of appraisal of what was done in such explanations; that what we want to know when we ask to have the action explained is in what way it was appropriate.103 Later he says, It is not necessary for the historian to show that the agent had reason for what he did; it is suf ficient for explanation to show that he had rea sons . But the element of appraisal remains in that what the historian declares to have been the agent's reasons must really be reasons (from the agent's point of view).104 But if rational explanation is characterized in this fashion, it surely suffers from the fact that it cannot be used to explain the entire spectrum of human actions, and it is especially defective in explaining abnormal actions. That is, it would not appear to be possible to explain irrational, 103Ibid., p. 261. 104Ibid., p. 263. 133 ill-judged, or ill-thought-out actions on this model. Hitler's actions late in the war may not, for example, be capable of being explained on this model if his thinking was sufficiently clouded by his obsession for world domina tion and his belief in the invulnerability of Nazi Germany. This defect then severely limits the scope of rational explanations. Yet it should be pointed out that although Dray does not deny that we could give nomological explana tions for defective as well as normal cases, he does not believe that this in any way damages his case against Hempel and the covering-law model. About such defective cases he claims, My claim is only that the criterion of rational appropriateness does function for actions not judged to be defective in various ways; and that it sometimes i£ found to be applicable. And I add that when we do employ this criterion, we cannot help certifying the agent's reasons as good ones, from his point of view, for doing what he did. That is what is meant by "following his argument" or "seeing the relevance of the con siderations he took into account."105 As against this notion of rational explanation as formulated by Dray, Hempel has argued, among other things, that it may provide good grounds for saying that the ap- * propriate thing for the agent to do under those circumstances was x, but it in no sense provides us with good grounds for -’William Dray, "The Historical Explanation of Actions Reconsidered," in Philosophy and History, op. cit., p. 113. 134 asserting that the agent actually did x.^06 It is quite clear then that from Hempel's viewpoint nothing can count as an adequate explanation unless it provides good grounds, deductive or inductive grounds, for asserting that the phe nomenon in question did in fact occur. This requirement is, as we will see,'*’^ behind Davidson's claim that there cannot be a logical relation between reasons and actions since if there were, the statement that S wants to do A would entail that S does A, which it obviously does not. Hempel's use of this principle is stronger yet since his claim is that not only will the statement that S wants to do A not entail that S does A, but the statement of S's relevant wants and beliefs plus Dray's principles of actions will not be sufficient to show that S does or did A. The requirement that any adequate explanation of a phenomenon succeed in showing that the statement asserting the occurrence of that phenomenon can be inferred (either inductively or deductively) from the explanans, i.e., that all legitimate explanations are arguments, no doubt is the * logical consequence of the thesis of the symmetry of ex planation and prediction together with the thesis that a 10^Carl G. Hempel, "Reasons and Covering Laws in Historical Explanation," in Philosophy and History, op. cit., p. 115. 107 See pp. 155-156. 135 prediction is an argument. The difference between scientific prediction and explanation is merely "of a pragmatic charac ter. "108 The only relevant difference between the two being, in seeking an explanation for a phenomenon we know that the phenomenon in question has occurred; whereas when we are predicting the occurrence of a phenomenon we do not know that it has occurred. "It may be said, therefore, that an explanation is not fully adequate unless its explanans, if taken account of in time, could have served as a basis for predicting the phenomenon under consideration.The sup posed symmetry of explanation and prediction is the source of Raziel Abelson's dubbing of Hempel's model of explanation as the "fallacy of misplaced proof.According to Abelson, 108 Carl G. Hempel and Paul Oppenheim, "The Logic of Explanation," in Readings in the Philosophy of Science, ed. by Herbert Feigl and May Brodbeck {New York: Appleton- Century-Crofts, Inc., 1953), pp. 322-323. 109 Ibid., p. 323. The thesis of the symmetry of pre diction aricl "explanation has been criticized by Michael Scriven in "Explanations, Predictions, and Laws," in Min nesota Studies in the Philosophy of Science, Vol. Ill: Scientific Explanations, Space, and Time, ed. by Herbert Feigl and Grover Maxwell (Minneapolis: University of Min nesota Press, 1962), pp. 170-230. ^°Raziel Abelson, "Cause and Reason in History," in Philosophy and History, op. cit., p. 172. 136 the covering-law model is simply a systematic way of proving what does not need to be proved, namely, that S did A. But we already know that S did A. This objection, however, cannot be thought of as destroying the covering-law model since even if, as is the case with historical explanation, we already know that the agent performed the act, it does not follow that an explanation is not in order or that the covering-law model cannot succeed in explaining why the act was performed. Likewise, knowing that Napoleon was defeated at Waterloo in 1815 does not preclude giving an explanation of why he was defeated. In fact, due to Napoleon's expertise at military tactics, it is just the sort of event which cries out for an explanation. As against Dray's notion of rational explanation Hempel claims that in order to get the required explanation of why the agent performed x, we shall have to add the law that "Any rational agent, when in a situation of kind C, will invariably (or with high probability) do x" together with the statement that the agent in question was a rational agent at the time.-^-1*^ Yet, in all fairness to Dray, Hempel's re quirement that all legitimate explanation must provide good ^^■Carl G. Hempel, "Reasons and Covering Laws in Histor ical Explanation," in Philosophy and History, op. cit., p. 155. 137 deductive or inductive grounds for asserting that the phenomenon in question did in fact occur may be interpreted as simply begging the question against Dray. If this should not be allowed, then it is obvious that the real point of contention between Dray and Hempel is exactly to what extent it can be shown that the law mentioned by Hempel is a genuine empirical generalization rather than an analytic statement about what we mean by a rational agent. Yet Hempel has offered some considerations which tend to show that the lav; in question cannot be considered to be purely analytic. First, no adequate general criteria of rationality have yet been formulated, which suggests that 112 there is no one clear sense of rationality in general. Secondly, there would seem to be a certain epistemic inter dependence of belief attributions and goal attributions. That is, as we have already seen, the ascription of an ob jective or goal to an agent implies something about his overt behavior only when it is taken in conjunction with the ascription of some appropriate beliefs. This inter dependence does not, however, make it impossible to determine an agent's objectives or beliefs from his overt behavior provided that we have some antecedent information about either 112Ibid., p. 154. 138 one of them^ Thus, provided that we have good antecedent grounds for the assumption that Brown believes that the ice is thin, we can infer from the fact that he keeps to the shallow areas around the edge that he wishes to avoid falling through the ice into deep water. And likewise, knowing that Brown desires not to fall through the ice into deep water, we may usually infer from his always keeping to the shallow areas around the edge that he believes that the ice is thin. This epistemological interdependence may create the impression that the law, "Any rational agent, when in a situation of kind C, will do x" is analytic rather than syn thetic. Yet this same epistemological interdependence of belief and goal attribution would seem to cast some doubt on the necessity of including in the explanation of an action the hypothesis that the agent in- question was rational at the time he performed the act. This is so because of the rather loose connection between the hypothesis and the overt behaviors which serve as tests of the relevant beliefs and objectives of the agent. That is, it would seem that the hypothesis of the agent's rationality is always trivially true because of the way in which v/e use the agent's behavior to determine his goals and beliefs. Thus, there is no need 1-^3Carl G. Hempel, Aspects of Scientific Explanation (New York: The Free Press, 1965), pT 415. 139 to suppose that if an agent's behavior fails to conform to a particular ascription of belief or objective, then we must search for some other belief or objective which would account for the behavior. For, it is not only conceivable, but it sometimes actually happens that we s£ick to our ascriptions of the agent's beliefs and objectives and in stead abandon the hypothesis of his rationality. There are perhaps innumerable ways in which an agent may fail to act rationally. He may, for example, simply fail to take certain relevant factors into account which he clearly believes and if they were taken into account, they would have resulted in a different course of action. He may likewise fail to notice certain aspects of his objective which, if they were taken into account, would again have called for a different course of action. And finally, even if he goes through a deliberate process of deliberation with all of the relevant features of the situation and his goals and beliefs clearly noted, there may still be some logical mistake in his calculation. Thus, insofar as Dray's rational explanation depends upon the hypothesis of the rationality of the agent, its scope is seriously limited, and the epistemological interdependence of belief and goal ascription does not imply the analyticity of the law, "Any rational agent, when in a situation of kind C, will do x." H^Ibid. , p. 476. 140 3.4. Reason Concepts as Dispositional Neither the failure of the supporters of rational explanation to provide adequate general criteria of ration ality/ nor the epistemic interdependence of belief and objective ascription are sufficient to show that the state ment "Any rational agent/ when in a situation of kind C, will do X" is synthetic. In order to make his claim more plausible, Hempel argues that the law in question must be considered to be an empirical generalization because it is capable of being analyzed dispositionally. In particular, his claim is that such concepts as rationality are . . . governed by large clusters of general state ments - they might be called symptom statements - which assign to the dispositional characteristic in question various typical manifestations or symp toms; each symptom being a particular manner in which a person who has the dispositional character istic will "respond to" or "act under" certain specific ("stimulus") conditions. Now what I want to argue primarily in Chapter V is that there is no good reason why a dispositional account of mo tives, beliefs, and rationality cannot be considered as at least part of any satisfactory account of human actions. Probably one of the major reasons for the reluctance on the part of many philosophers to accept the covering-law 1^5Carl G. Hempel, "Reasons and Covering Laws in Historical Explanation," in Philosophy and History, op. cit., p. 156. 1 141 model of explanation is that the covering-law theorist's attempts at providing strict nomological explanations of human actions are either nonexistent, or redundant and platitudinous, or simply irrelevant. Hempel, for example, offers the following covering-law explanation of why the dustbowl farmers migrated to California in "The Function of General Laws in History": Drought and sandstorm render a farming existence precarious. California seemed to offer better living conditions, and populations tend to migrate toward regions which offer better living conditions. And in The Structure of Science, Ernest Nagel offers the following probabilistic explanation of why Cassius plotted against Caesar: . . . In ancient Rome the relative frequency (or probability) was high (e.g., greater than one- half) that an individual belonging to the upper strata of society and possessed by great hatred of tyranny would plot the death of men who were in a position to secure tyrannical power. Cassius was such a Roman and Caesar such a potential tyrant. Hence, though it does not follow that Cassius plotted the death of Caesar, it is highly probable that he did do so.^l^ 116Carl G. Hempel, "The Function of General Laws in History," in Readings in Philosophical Analysis, op. cit., p. 464. 117 Ernest Nagel, The Structure of Science: Problems in the Logic of Scientific Explanation (New York: Harcourt, Brace & World, Inc., 1961), pp. 22-23. 142 Wow even though Hempel admits that the universal hypothesis that populations will tend to migrate to regions which offer better living conditions is not very precise and that we are in no position to state the well-confirmed general law involved, and Nagel admits that his explanation depends on a highly statistical hypothesis, their failure to come up with more precise laws has caused some to wonder if there are any more precise laws to be discovered. And these ten dency statements have struck some philosophers as nothing more than arbitrary generalizations of single cases, clev erly contrived so that no explanation could possibly escape being an instance of the covering-law model of explanation. But in spite of the relatively artificial appearance of these proposed explanatory hypotheses, it is highly question able to what extent Dray can show that his principles of action are evaluative in a way in which empirical generali zations are not, or that the action theorist can show in what way reasons explain, actions, which is significantly different from the way in which laws are said to explain actions. Un fortunately, Dray himself is never terribly specific about the exact sense in which his principles of action are evalua tive. Nowhere does he attempt to provide a clear elucidation of the way in which citing the reasons, beliefs, and princi ples of action explain an agent’s actions. The closest he 143 comes to it is to say that they show that what the agent did was the appropriate thing to do from his own point of view. But this promises to remain hopelessly obscure unless some analysis of elucidation of the way in which citing the agent's reasons and beliefs can be said to explain his action can be given, and we have already seen in Chapter II that Melden's and Louch's attempts at elucidating this sense have proven unsatisfactory. 3.5. The "Logical Connection" Argument There have been two major considerations which have led philosophers to argue that reasons are not causes: (1) Rea sons are known in a different way than causes. (2) There is a logical relation between reasons and the actions for which they are reasons, but not between causes and effects. I shall begin with the latter of these. The so-called "logical connection" argument, although frequently employed by Melden and others,has been criti cized by a number of contemporary philosophers. In outline, the argument purports to show that motives, intentions, and other reasons cannot be causes because they lack at least one of the essential features of causes. More specifically, 1 18 R.S. Peters, The Concept of Motivation (2nd ed.: New York: Humanities Press, 1960), p. 113. it claims that motives, intentions, and other reasons can only be understood or identified by reference to the actions of which they are the reasons. Hence, a motive is not just a motive pure and simple, but is always a motive for some action or other, and an intention is not just an intention, but is always an intention to do or omit doing some action or other. In other words, the fact of the matter is that when we come to the problem of describing our motives, we find that we can only describe them in terms of actions like poisoning our neighbor's dog, investing in a particular stock or returning an overdue book to the library. The same thing seems to hold true for our descriptions of our intentions, for instance, we intend to buy a first edition, pay off a debt, or become a world renowned expert on winepresses. Thus according to the argument, there seems to be absolutely no way a reason can be described or referred to except by includ ing in its description a description of the action of which it is the reason. But it is just this feature of descriptive dependence which the cause-effeet relation seems to lack. That is, the reason-action relation seems to violate the Humean criterion that the event identified as the cause must be descriptively independent of the event which is the effect of that cause. Therefore, it follows that reasons cannot be causes, and that the relation between reasons and actions must be quite different from the relationship existing between causes and effects. 145 Yet in spite of its apparent popularity, this argument has been severely criticised, and I believe correctly, by a number of philosophers in recent philosophy.-^9 Probably one of the most unsympathetic critic3;has been Donald Davidson. Professor Davidson is correct in pointing out that there are at least two different versions of the argument.I20 The first form of the argument seems to hinge upon the notion of logical distinctness. In short, this form of the argument claims that although causes and effects are logically distinct, rea sons and the actions they explain are not. This way of pre senting the argument is, however, potentially misleading. It may tempt one to think that the argument is asserting that objects, processes, properties, and events which are the sorts of things which can be causally related are logically dis tinct, while reasons and actions are not. This would be a serious mistake since such an interpretation might tempt one to argue that since only meanings or propositions, not ob jects, processes, properties, or events, are the proper sorts of things about which it makes sense either to affirm or deny 119 Donald Davidson, "Actions, Reasons, and Causes," o p . cit., p. 75; Alvin I Goldman, A Theory of Human Action (Englewood Cliffs, New Jersey: Printlce Hall, Inc.), 1970; A.C. Ewing, "May Can-Statements Be Analysed Deterministic-' ; ally?", Proceeding of the Aristotelian Society, LXIV (1964), pp. 171-172; W.D. Gean, "Reasons and Causes,'1 The Review of Metaphysics, XIX (1966). l20Donald Davidson, "Actions, Reasons, and Causes," op. cit., p. 75. 146 are logically distinct, the "logical connection" argument is a prime example of a piece of nonsense. But it is clear that, in spite of Davidson's and Melden's language here to the contrary, they do not intend to be asserting anything of this sort. What Melden means to be asserting is that whereas the concept or idea of reasons are not logically distinct from the concept or idea of the actions for which they are reasons, the concept or idea of a cause is logically distinct from the concept or idea of its effect. To avoid this rather simple misinterpretation, the first version of the "logical connection" argument should be reformulated in the following manner: cause events (objects) can be cor rectly described in terms which do not appear in the cor rect description of the alleged effect events (objects); but reasons can only be correctly described in terms which appear in the correct description of the appropriate actions; therefore, reasons cannot be causes of actions.^ 1 The second version of the argument asserts that to give a reason for an agent's action is really to redescribe it, and thus, there are not two events being described but only 12lThis is not the first instance in which the notion of descriptive independence has been used by a philosopher. It was used, for instance, by O.K. Bousma in rejecting G.E. Moore's arguments for the existence of sense-data. See Part II of his essay, "Moore's Theory of Sense-Data" in The Phil osophy of G.E. Moore, ed. by P.A. Schilpp (Evanston^ Illinois: Northwestern University Press, 194 2). 147 one. To explain why a driver extends his arm out of the window on approaching a turn by saying that he wants to reach the drug store on the adjacent street is, therefore, to redescribe his action of raising and extending his arm out of the car window as the action of signalling a turn. In this case there are not two events or sets of events which are being described as "extending his arm out of the car window" and "signalling" respectively, but only one event or set of events. But to give a cause for an event is not to merely redescribe it, for otherwise, causes and the events of which they are causes would not be logically distinct events. It follows that reasons cannot be causes. Unfortunately, it is not always easy to tell exactly which one of these versions Davidson is criticizing at any one time. If I understand him correctly, Davidson has two major objections against the first version of the argument. The first consists in arguing that causes and effects are not contingently related. There is something very odd, Davidson tells us, about the notion that causal relations are empirical rather than logical, and according to Davidson, its oddness is a function of the fact that not every true causal statement is empirical. "For suppose," Davidson says, "'A caused B 1 is true. Then the cause of B=A; so, sub stituting, we have 'The cause of B caused B,' which is 148 analytic. "122 This argument is, however, hardly substantial enough to refute the "logical connection” argument. Ap parently, Davidson thinks that any statement is a causal statement as long as it contains the words "cause" or "ef fect." And in this sense, the statement, "The cause of B caused B" is indeed an analytic causal statement. This is, of course, absurd. Far more is required for a statement to be a statement which asserts a causal relation than that it simply contain the words "cause" or "effect." It is ob vious that those philosophers who claim that causal rela tions are empirical do not mean to assert that all true causal statements are empirical. Consider "all effects are caused."123 Nor ^oes Davidson's argument show that the statement, "King Charles the First had his head cut off and did not die," and similar causal statements, are not logical ly possible. Davidson's criticism is also faulty on other grounds. For there are clearly two alternative ways in which the ex pression "the cause of B" may be understood. The first sense in which this expression may be read is "Whatever caused B." On this reading the resulting sentence would 122Donald Davidson, "Actions, Reasons, and Causes," op. cit., p. 75. l22Cf. Raziel Abelson, "Doing, Causing and Causing to Do," Journal of Philosophy, LXVI (1969), pp. 178-192. 149 read "Whatever caused B caused B," which is a logical truth comparable with "What caused B is what caused B." But Davidson1s argument clearly depends upon treating the ex pression "the cause of B" as a purely referring expression, or else it could not be used in the identity statement, "The cause of B=A.tl3-24 A purely referential position must be subject to the substitution of identity,and given a true statement of identity, one of its terms may be sub stituted for the other in any true statement and the result will be true. Hence, if the function of the expression "The cause of B" is solely to refer, then the resulting statement, "the cause of B caused B" is not analytic since it says that some one event (no matter how referred to) was the cause of B and this is not trivially true. Davidson's second argument is much more serious. This argument proceeds in terns of Davidson's example of the man who flips a switch, turns on the light, illuminates the room, and unknown to him, alerts a prowler to the fact of •^■2^Cf. Willard van Orman Quine, Word and Object, (Cam bridge, Massachusetts: The M.I.T. Press, 1960) pp. 142 ff. 125Cf. Willard van Orman Quine, From a Logical Point of View, 2nd ed., (New York: Harper and Row, Publishers, Inc., 1961), p. 139. 150 his presence. In reference to this example, Davidson argues, The example serves also to refute the claim that we cannot describe the action without using words that link it to the alleged cause. Here the action is to be explained under the description: "my flipping the switch," and the alleged cause is "my wanting to turn on the light." What pos sible logical relation is supposed to hold be tween these phrases? It seems more plausible to urge a logical link between "my turning on the light," and "my wanting to turn on the light," but even here the link turned out, on inspection, to be grammatical rather than logical.126 As I see it, there are at least two things which are at issue here. First, to what extent is Davidson justified in citing as a reason for the act of flipping the switch my wanting to turn on the light? And second, to what extent is Davidson justified in claiming that the relation ship between my reason, wanting to turn on the light, and the action under the description "my turning on the light" is more grammatical than logical? To what extent is Davidson justified in citing as a reason for my act of flipping the switch my wanting to turn on the light? If the sentence, "I wanted to turn on the light," actually does represent my reason for flipping the light switch, then it would appear that Davidson has suc cessfully shown that a reason can be described and identi fied independently of the action for which it is a reason. 126Donald Davidson, "Actions, Reasons, and Causes," op. cit., p. 75. 151 Davidson, however, does not present any positive argument here. All he does is issue the challenge to find a logical link between the sentences, "I wanted to turn on the light,” and "I flipped the switch.” Myles Brand, however, does Davidson one better in providing a positive account of why we cannot say that there is a logical link here. Thus, Brand argues, The premise that a reason and its resultant behavior are not contingently related is unjustified. Sup posing that reasons are causes, it is possible never theless to identify and describe them without referring to the resultant behavior. For a reason can be iden tified and described by referring to its goal. Sup pose that I have a motive (reason) for signalling the chairman. Suppose also that just as I am about to signal, someone or something prevents me. I could nonetheless identify and describe my motive, not by the resultant behavior - which did not occur - but rather by the goal of having been recognized by the chairman. Sometimes in fact we identify or describe a motive by referring to the resultant behavior. We can do this, however, only when the goal and the resultant behavior coincide, that is, when we achieve our goal. Again, we recognize, identify, and describe motives (reasons) not by their actual outcomes, but rather by their expected or proposed outcomes. The reason for claiming that reasons and their resultant behavior are not contingently related is, then un founded. To what extent is Davidson justified in claiming that the relation between ”my wanting to turn on the light” and "my turning on the light” is more grammatical than logical? 127Myies Brand, The Nature of Human Action, op. cit., p. 12. 152 Davidson attempts to justify this claim in the following way: We may be taken in by the verbal parallel between "I turned on the light11 and "I wanted to turn on the light." The first clearly refers to a parti cular event, so we conclude that the second has this same event as its object. Of course it is obvious that the event of my turning on the light can't be referred to in the same way by both sen tences, since the existence of the event is re quired by the truth of "I turned on the light" but not by the truth of "I wanted to turn on the light." If the reference were the same in both cases, the second sentence would entail the first; but in fact the sentences are logically independent. What is less obvious, at least until we attend to it, is that the event whose occurrence makes "I turned on the light" true cannot be called the object, how ever intentional, of "I wanted to turn on the light." If I turned on the light, then I must have done it at a precise moment, in a particular way - every detail is fixed. But it makes no sense to demand that my want be directed at an action performed at any one moment or done in some unique manner. Any one of an indefinitely large number of actions would satisfy the want, and can be considered equally eligible as its object.128 A careful examination of this argument reveals that it is really composed of four different parts. First, the claim that it makes no sense to demand that the reason for an action be precisely determined in the way that the action is, if it occurs. Second, the claim that since the event required by the truth of "I turned on the light" is not required by the truth of the sentence, "I wanted to turn on the light," these two sentences cannot refer to this event in the same ^2®Donald Davidson, ''Actions, Reasons; and Causes," op. cit., p. 69. 153 way. Third, the claim that if these two sentences actually did refer to the same thing, then "I wanted to turn on the light” would entail "I turned on the light." And finally, the claim that the event which makes "I turned on the light" true cannot be the intentional object of the sentence, "I wanted to turn on the light" since that event is precisely determined in a way that the want is not. Now Davidson is quite right in claiming that it makes no sense to demand that the reason for an action be pre cisely determined in the way that the action is, if it occurs. And this is so in spite of the fact that oc casionally a want or desire may be for an action which is performed at some precise time or in some precise manner, e.g., I want to snap the picture at the precise moment that he gets hit with the water bag, I want to sink the seven ball in the corner pocket by bouncing it off three cushions. The second claim is so obvious that it deserves no further comment. However, the import of the third claim is debatable and I shall put off discussion of it until later. The fourth claim, I believe, is justified. For the intentional object of a reason (e.g., a desire) is something like a concept, rather than an actual act or object. The intentional object of my desire to climb Mount Everest is the concept of my climbing Mount Everest, and not the nonexistent act of my climbing Mount Everest. Likewise, my desire for a 154 cold glass of water has as its intentional object the concept of a cold glass of water, or my drinking a cold glass of water, and not the actual cold glass of water, or the actual act of drinking the water. There is indeed some evidence that Melden does confuse the intentional object of a desire with the actual thing or action desired. Hence, he states, It must be possible to characterize that internal impression without invoking any reference to the so-called object of the desire, no less than the action that consists either in getting or in trying to get that object. But as a desire, no account is intelligible that does not refer us to the thing desired.*29 Nov? there can be little question about whether or not the concept of the reason is logically connected with the concept of the object or action for which it is the reason. That is, there can be little question of whether the concept of a reason and its intentional object are logically related or not. But although it will be granted that a desire is logically related to its intentional object, there is still some .question about whether the desire itself is logically connected with the actual object of action. Melden, Free Action, op. cit. , p. 114. 155 Davidson's third claim, however, has brought a criti- 130 cism from Raziel Abelson. According to Abelson, Davidson and some other theorists have mistakenly supposed that when it is claimed that there is a logical relation between reasons and actions that what this means is that if an agent has a reason for performing some action, then the agent must perform that action, i.e., the statement, "S wants to do A" must entail that S does A. But this is, he claims, absurd since the logical bond between a reason and the action for which it is the reason is not one of entailment. The statement that S wants to turn on the light does not, by itself, entail that S will turn on the light, or even that he will flip the switch. The "logical con nection" argument, Abelson claims, does not claim that the statement of a man's reason for doing something logically entails that he will do it is the sense in which there is a logical connection between reasons and actions. What Abelson wishes to substitute for this strict entailment is the v/eaker thesis that S's reason for doing something entails that he will do it, ceteris paribus. 3-31 ^•3®Raziel Abelson, "Doing, Causing, and Causing to Do," The Journal of Philosophy, LXVI (1969) pp. 183-184. I31Ibid., pp. 183-184. 156 This criticism of Abelson is faulty, however. For one thing, we have already noted that both a belief ascription and a want ascription are necessary for even a minimally adequate explanation of a human action. So far as X know, there are no general statements relating a want considered alone with any action. Thus, no statement to the effect that "S wants to do A” by itself entails that S will do A. Secondly, so far as I know, neither Abelson nor anyone else has succeeded in clearly specifying the conditions under which the statement of an agent's reason for doing something entails that he will do it. Such uses of ceteris paribus clauses do not serve to enlighten us about the precise con ditions under which a statement of an agent’s reason for doing something entails that he will do it. The most plausible suggestion and one which X will present in Chapter V is that this ceteris paribus clause may in some cases be filled in with a statement of the agent's relevant beliefs and that beliefs, in conjunction with wants, cause actions. In arguing ggainst the second version of the "logical connection" argument, Davidson says, And there is no denying that this is true: when we explain an action, by giving the reason, we do redescribe the action: redescribing the action gives the action a place in a pattern, and in this way the action is explained. Here it is tempting to draw two conclusions that do not follow. First, we can't infer, from the fact that giving reasons merely redescribes the action and that causes are separate from effects, that therefore reasons are 157 not causes. Reasons, being beliefs and at titudes, are certainly not identical with actions; but, more important, events are often redescribed in terms of their causes. Suppose someone was injured. We could re describe this event "in terms of a cause" by saying he was injured. Second, it is an error to think that, because placing the action in a larger pattern explains it, therefore we now understand the sort of explanation involved. 3 I agree with Davidson that explaining an action by giving the reason and thus placing it in a larger pattern does not show that reasons are not causes. I also agree that by noting this feature of reason-giving we are not one bit closer to understanding what sort of explanation has been given. The latter is particularly evident from the general failure of action theorists to make this sense of explanation clear. Yet, the fact that events can frequently be described in terms of their causes is not particularly damaging to the argument. For the unique feature of reasons, it has been claimed, is that they can never be identified or correctly described independently of the actions for which they are reasons, while causes can always be identified or described independently of their effects, although it is not the case that this is always done. Nevertheless, the claim that reasons cannot be described or identified independently of 132Donald Davidson, "Actions, Reasons, and Causes," op. cit., p. 72. 158 actions is simply false. Reasons can be described or identified by means of definite descriptions, e.g., "The desire that the agent had at time t^.,! Thus, there is little plausibility in the claim that reasons cannot be causes because reasons are logically connected to actions, while causes cannot be logically connected to their effects. 3.6. Knowing Our Reasons Versus Knowing Causes It has sometimes been argued that reasons are not causes because there is a relevant difference between the way in which a person comes to know a cause of an event and the way in which a person comes to know his reasons for acting. Such an argument seems to be implicit in G.E.M. Anscombe's claim that reasons, or what she calls "mental causes," are in the class of things which are known without observation. It is very difficult to tell exactly what Miss Anscombe has in mind as the meaning of the phrase "known without observation." This ambiguity is evident also in Donald Davidson's attempt to refute the argument. Professor Davidson characterizes the argument as follows: It is said that the kind of knowledge one has of one's own reasons in acting is not compatible with the existence of a causal relation between reasons 133(3^.M. Anscombe, "Intention," in The Philosophy of Action, ed. by Alan R. White, op. cit., p. 145. For a similar argument, but without the notion of "mental causes," see A.I. Melden, Free Action, op. cit., pp. 138-139. 159 and actions: A person knows his own intentions in acting infallibly, without induction or ob servation, and no ordinary causal relation can be known in this way.1^4 Prom this passage it is apparent that Professor Davidson interprets the argument as claiming at least two different things: first, that a person's knowledge of his own rea sons for acting is infallible, and secondly, that the know ledge that a person has of his own reasons for acting is neither inductive nor based upon observation. Regarding the first of these claims, I think that Professor Davidson has correctly shown that it is perfectly possible for a person to err about his own reasons for acting, especially when the agent has two reasons for acting, one which agrees with the agent's ideal picture of himself and the other which does not agree with that picture. Hence, to borrow Davidson's example, a person who performs a mercy killing may really be motivated by the desire to be rid of the person rather than the desire to relieve him of his suffering. All of this is perfectly correct, but one wonders why it was necessary. For so far as I can determine, none of the major proponents of the view that reasons cannot be causes have ever claimed that we have an infallible know ledge of our own reasons for acting. 134Donald Davidson, "Actions, Reasons, and Causes," in _The Nature , of Human Action, ed. by Myles Brand, op. cit., p. 160 It would appear that Professor Davidson essentially agrees that usually the knowledge that a person has of his own reasons for acting is neither inductive nor based upon observation since he claims that though on occasion we will accept private and public evidence as showing that we were wrong about our reasons, usually we have no evidence and make no observations. And he also agrees that this shows that our own knowledge of our own reasons for acting is not inductive since a necessary condition for induction is evi dence. What he does not agree with is the further impli cation that what this shows is that the knowledge that we have of our own reasons for acting is not causal. In arguing against this inference, Davidson claims that there are other ways of coming to know that a singular causal statement is true besides.by induction. Hence, he argues, Causal laws differ from true but nonlawlike gen- alizations in that their instances confirm them; induction is, therefore, certainly a good way to learn the truth of a law. It does not follow that it is the only way to learn the truth of a law. In any case, in order to know that a singular causal statement is true, it is not necessary to know the truth of a law; it is necessary only to know that some law covering the event at hand exists. And it is far from evident that induction, and induction alone, yields the knowledge that a causal law satis fying certain conditions exists. Or, to put it dif ferently, one case is often enough, as Hume admitted, to persuade us that a law exists, and this amounts 161 to saying that we are persuaded, without direct inductive evidence, that a causal relation exists. Davidson's attempt to gain support from Hume for his argument strikes me as a bit curious, however, for even though Hume is frequently cited as showing that it is not necessary to have more than one instance of an alleged causal relation between two events to convince us that there must be some causal law which relates these two events, he did not claim that we were justified in believing this. In dis cussing this kind of case, Hume says, 'Tis certain, that not only in philosophy, but even in common life, we may attain the knowledge of a particular cause merely by one experiment, provided it be made with judgment, and after a careful removal of all foreign and superfluous circumstances. Now as after one experiment of this kind, the mind, upon the appearance either of the cause or the effect, can draw an inference con cerning the existence of its correlative; and as a habit can never be acquired merely by one in stance; it may be thought, that belief cannot in this case be esteem'd the effect of custom. But this difficulty will vanish, if we consider, that tho' we are here suppos'd to have had only one experiment of a particular effect, yet we have millions to convince us of this principle; that like objects, plac'd in like circumstances, will always produce like effects, and as this principle has establish'd itself by a sufficient custom, it bestows an evidence and firmness on any opinion, to which it can be apply'd. The connexion of the ideas is not habitual after one experiment; but •*-33Ibid., p. 78. 162 this connexion is comprehended under another principle, that is habitual; which brings us back to our hypothesis.136 Now it is important to note here that Hume is attempting to explain why it is that we have a tendency to suppose that a causal law exists connecting two events which we have seen occurring together only once. He is not attempting to show that such a tendency is by any means justified. Hence, contrary to Davidson, showing that we have a tendency to believe that there is causal lav; connecting two events which we have experienced only once, is not showing that we are justified in doing so. Another curious feature of Davidson's argument is the claim that we can know that a causal law exists without direct inductive evidence that a causal relation exists. Exactly why is this qualifying phrase necessary? Surely some form of induction is at work if we conclude that two events are causally related because they are similar to other events which we know are causally related. In other words, in order to conclude that two events which we have exper ienced for the first time together are causally related, we should have to show that the alleged cause was similar to events which we already know to be causes, and that the 136David Hume, A Treatise of Human Nature; Being An Attempt to Introduce the Experimental Method of Reasoning Into Moral Subjects’! ed. by L.A. Selby-Bigge (Oxford, England: Oxford University Press, 1888), pp. 104-105. 163 alleged effect was similar to events which we already know to be effects of these said causes. In this respect, Davidson's argument begs the question since it merely assumes, without argument, that reasons are like events which we al ready know to be causes. A final curiosity about Davidson's argument is how we can be said to know that a lav; exists which relates two events without knowing the truth of that lav;. Perhaps what Professor Davidson has in mind here is the case in which a researcher can observe a certain constant conjunction be tween events of a certain sort and then set out discovering some formula which will best express that correlation. But this situation is better described as a situation in which the researcher suspects that there is some law which covers the events in question, not a situation in which he knows that there is some law. To say, as Davidson does, that he knows that there is some lav; which covers the events in question, but that he does not know the truth of that law, is dangerously misleading. But even if Davidson is not successful in refuting this argument against treating reasons as causes, it is doubtful whether the argument, as it stands, is sufficient to show that reasons are not causes. For first, as was already pointed out, the phenomenon of repression suggests that be havior may be motivated in some cases by desires about which 164 the agent has no knowledge, immediate or otherwise. To argue that a repressed or unconscious desire is not a desire in the usual sense of the term and hence that it is essential to having a desire that the agent know he has it, is merely to beg the question. And even if it is admitted that the notion of a repressed or unconscious desire is not the usual meaning attached to the concept, this does nothing to show that the notion of a repressed on unconscious desire is meaningless or self-contradictory. And secondly, there seems to be no very good argument to show that those cases in which we do know our reasons are significantly different from those cases in which we know causes or causal conditions. For it is not in the least clear exactly what is being contrasted when it is claimed that reasons are known without observations while causes are not. The major part of this unclarity hinges on the obscurity of the very notion of observation itself. For if what is meant is that our own reasons are things of which we are immediately conscious, then it is difficult to see how they could be distinguished from pains, idle thoughts, perceived colors, shapes, sounds, and so on.. Furthermore, if this is the sense in which reasons are known without obser vation, it is difficult to imagine what it is being contrasted with. For what would count as being mediately.consciousness? Therefore, the argument that reasons are not causes because they are known without observation is inconclusive. One initially plausible suggestion is that the relevant distinction between the way in which we know our own reasons for acting and the way in which we know causes is that while our knowledge of causes is inferential, our knowledge of our own reasons for acting is not. But such a suggestion is insufficient to establish that there is any relevant dif ference betv/een the way in which we come to know our reasons and the way in which we come to know causes or physical ob jects. For there is no requirement in either case that the inference be in any sense a conscious inference. Thus, I know that I have a pen in my hand at this moment because I use certain subtle variations in color shades and patterns and tactile variations, but I am unable to give a verbal account of these variations, nor am I aware of making any conscious inference from them and likewise there seems to be nothing to prevent me from arriving at a knowledge of my own reasons in a similar fashion. For it is perfectly possible that when I know I desire to hear music, for example, I read this off from subtle feelings and tensions which occur in me whenever something in the environment produces it, whenever, for example, the prospect is mentioned by others or myself. And in this case, I am aware of no conscious inference on my part nor am I able to give an account of how I do this. 166 This sense of knowing something without observation seems to be what is behind the claim that we do not ordinarily see what people do in terms of mere bodily movements, but rather in terms of actions. Melden, for example, claims, . . . For there is a difference between placing coloured pieces of paper in the palm of a person's hand and making a payment to a grocer for food received; yet paying a grocer is not less an action, something one can observe taking place, than moving a piece of coloured paper from here to there by executing the required and in itself complicated bodily action.13? and again later In most cases there is no interpretation at all, no reflection, consideration or decision, no exercise of judgment, no room for hesitation or doubt. We simply see a person raising his arm ju^fe as we read off what we find on the printed page. But such claims cannot be interpreted as showing that there is any relevant difference between the way in' which we come 137a .I. Melden, Free Action, op. cit., p. 178. (Italics mine.) 138Ibid., pp. 186-187. (Italics mine.) Melden is by no means alone in this claim. P.F. Strawson, for example, writes, "Among the things that we observe, as opposed to the things we know without observation, are the movements of bodies similar to that about which we have knowledge not based on observation . . . we understand them, we interpret them, only by seeing them as elements in just such plans or schemes of action as those of which we know the present course and future development without observation of the relevant present movements. But this is to say that we see such movements as actions . . P.F. Strawson, Individuals, (Garden City, New York: Anchor Books, Doubleday and Company, Inc., 1963), pp. 108-109. 167 to a knowledge of bodily movements or causes and human actions and our own reasons for acting. Thus, for example, although what is "said’' by a sentence is something dif ferent from the curiously shaped black marks on a white background, it does not follow that that something else is not, as it is in most cases, a substantial proportion of unconscious interpretation. If, on the other hand, the intended contrast between those things which are known without observation and those things which are known with observation is supposed to be a contrast between those things which are known without re peated observation and those things which are known only after repeated observation, then this contrast would, at most, only create problems for a philosopher who was al ready committed to a notion of causation as essentially involving constant conjunction. For if we construe the causal relation in the Humean fashion, then it would appear that it can only hold between two distinct events. But it would appear that in many cases in which an agent acts from a motive and is even conscious of his motive in acting, his action is not preceded by any psychological event which can be identified as his motive in acting. There is no require ment that the agent, even, if he is conscious of his motive in acting, must actually formulate it even to himself, and 168 if the motive is something over and above the agent's behavior it may simply consist in his ability to say what it is. But none of this precludes the possibility that reasons are causes. For first, there is absolutely no reason why a philosopher must be committed to a Humean notion of causation. The Humean conception may simply be too restrictive to account for the ordinary and scientific senses of cause. It may turn out that invariable sequence is re quired ideally for the full-fledged acceptance of a causal relation or causal law, but it is clearly not an absolute requirement for either the ordinary or even most scientific uses of causal language. One plausible suggestion for ex tending the notion of cause so that it could include these other cases and still leave open the possibility of realiz ing the ideal is to regard causal relations as holding between facts, where "facts" is understood as referring to what is expressed by any true proposition.139 This procedure would have the dual advantage of extending the scope of causal relations so as to include most ordinary and scien tific uses of causal language and at the same time allow Ayer, Man As a Subject for Science, in Readings in Ethical Theory, ed. by Wilfred Sellars and John Hospers (2nd ed.; New York: Appleton-Century-Crofts, 1970), p. 680. 169 for the future realization of the ideal requirement of invariable sequence when we are able to translate these statements about facts into statements about events. In deed, it is just the extreme ambiguity of the term "cause" and the possibility of applying it to cases in which there is something less than invariable sequence which accounts for what little plausibility the arguments against identi fying reasons with causes have. And furthermore, regardless of what the correct analysis of the causal relation is, it is not altogether clear that even a philosopher who is committed to the notion that constant conjunction is an essential feature of the causal relation would be extremely troubled by the present argument. For the argument as it stands does nothing to show that desires, wants, motives, and so on are not invariably connected with occurrences of certain characteristic responses, feelings, tensions, and the like, nor does it show that we do not come to a knowl edge of our own reasons for acting by an unconscious inference from such responses, feelings, etc., which may be invariably connected with them. It may be extremely difficult to produce any specific response, feeling, or ten sion which is always and invariably conjoined with a particu lar desire, want, or motive; but there is nevertheless no good reason for denying that this could be done. 170 In addition to these difficulties with the argument, the argument that reasons are not causes because they are known in a different way from causes is clearly fallacious. Its fallaciousness springs from the fact that modal contexts are referentially opaque. Thus, the argument in question may be formulated as follows: (1) Necessarily reasons are not known by observation. (2) Necessarily causes are known by observation. (3) Therefore, necessarily reasons causes. This argument, however, has exactly the same form as the following invalid argument: (1) Necessarily 9 > 4. (2) Necessarily the number of major planets = 9. (3) Necessarily the number of major planets > 4. For, even granted that premises (1) and (2) are true, (3) is clearly false. This clearly shows that the position of '9' in (1) is not purely referential and that the necessity operator is opaque. 3.7. Two Distinct Approaches to the Problem of Human Actions A large part of the confusions surrounding these two arguments are amplified by the general vagueness of the concept of a human action. For it is by no means clear 1^®Willard van Orman Quine, Word and Object (Cambridge, Massachusetts: The M.I.T. Press, 19 60), p. 197. 171 that what one philosopher would classify as an action another philosopher will also, or even that what one and the same philosopher would classify as an action will neces sarily have anything in common with the other things that he classifies as actions. One of the reasons there is so much confusion about the exact nature of a human action, I want to suggest, is that many action theorists, Melden, for instance, when they come to consider the nature of human actions, approach the problem from two distinct directions. The first way of approaching the problem is to ask, "What is the difference between 'my raising my arm' and 'my arm rising'?" The second and quite distinct way of approach ing the problem is to ask a question like, "What is the difference between 'my extending my arm* and 'my signalling for a left-hand turn'?" But it is not clear that what is being contrasted in these two approaches is in any sense the same thing. If the problem is approached as the problem of dis tinguishing "my raising my arm" and "my arm rising," then there are at least two contrasts which may be made. First, we may be contrasting my raising my arm as a result of wanting to raise it and my arm rising as the result of some unnatural internal "compulsion," e.g., the movement of my arm as the result of some muscle spasm, nerve disorder, or1 some other physiological disorder. Secondly, we may be 172 contrasting my raising my arm and my arm rising as the result of some external force, e.g., a sudden gust of wind forcing my arm up, or someone grasping my arm and forcing it up. But if, on the other hand, we appraoch the problem of distinguishing between "my raising my arm" and "my signalling for a left-hand turn," then there is a different set of contrasts involved. Here, the contrast involved may be a contrast between those behaviors for which there are an appropriate set of rules and practices and those for 1 a «1 which there are no appropriate rules or practices. Hence, the intended contrast may in this case be between that class of things which we do, such as signalling, point ing, getting married, graduating, answering the telephone, and so on, which depend for their intelligibility on the existence of appropriate practices and conventions; and that class of things we do such as moving my right index finger, raising my arm, and so on, which do not depend for their intelligibility on the existence of appropriate prac tices and conventions. The contrast here is strictly different from the former contrast. In the former case, the contrast was between bodily movements and bodily actions, while in this latter case the contrast is between different actions. CHAPTER IV ACT DESCRIPTIONS AND ACT GENERATION In Chapter III, I explored the question of what con stitutes a proper explanation of human actions, arguing that although there is perhaps a sense in which reasons are ordinarily thought of as explaining actions which is not the sense in which causes or laws explain events, this sense is extremely muddled and does not rule out the possibility of giving nomological explanations, causal or otherwise, for human actions. First, it was pointed out that many action theorists simply ignore the essen tial role that an agent's beliefs play in explaining his actions. Next, the question of whether Dray's so- called "principles of behavior" or "principles of ac tion" are primarily empirical generalizations or evalu ative principles was considered and it was argued that not only are such rational explanations extremely limited in scope, but also the sense in which they could be said to explain actions which is significantly different from the sense in which laws or empirical generalizations ex plain events was never clearly specified. For it is by no means clear how such principles of action together with a statement of the agent's reasons for acting can show that what the agent did was the appropriate thing 173 174 for him to do even from his own point of view. Next, I examined and rejected the two major considerations or arguments which have been advanced to show that there is a supposed logical gap between reasons and causes, viz., the so-called “logical connection" argument and the argument that reasons cannot be causes because they are known in a radically different way than causes. In regard to the so-called "logical connection" argument it was argued that although Donald Davidson's attempt to refute the argument by arguing that causes and effects are not contingently related failed, his claim that the supposed logical connection between reasons and causes is more grammatical than logical seems to be justified. For, in the first place, we can identify and describe reasons not in terms of their actual out comes, but in terms of their expected or proposed outcomes. And, in the second place, there is no reason to suppose that a reason for an action is precisely determined in the way that the resulting action is. Philosophers who suppose that there is a logical connection between reasons and actions have simply confused the intentional object of a reason with some actual existent thing of action. Next, Raziel Abelson’s attempt to save the claim that there is a logical connection between reasons and actions provided that certain ceteris paribus clauses are added was found to be in itself unenlightening, and it was suggested that the only plausible candidate for such ceteris paribus clauses is a statement of the agent's relevant beliefs which, in conjunction with certain laws, would entail that the agent perform a said action. The argument that reasons are known in a completely different way from the way in which causes are known and hence reasons cannot be causes also was examined and found to be faulty. For although some of Davidson's attempts at refuting this argument are not successful, the phe nomenon of repression and the intrinsic vagueness involved in the notion of things which are known v/ithout obser vation serve to weaken the argument to the point to which it cannot be considered to be a very plausible reason for distinguishing reasons from causes. And fi nally, it was argued that some philosophers have been mistakenly led to suppose that the defining character istic of all action is that they are bodily movements in certain circumstances in which conventions and prac tices apply from a failure to see that there are not one but several contrasts marked by the supposed distinction between human actions and bodily movements. In this chapter I shall show how this confusion can be avoided by developing a more refined method of describing and talking about actions. However, in order to carry out this proposal it will be necessary to draw a few vital preliminary distinctions. 176 4.1 Action Descriptions Problems associated with the question of what con stitutes a correct description of a human action are quite possibly the source of the greatest misunder standings. and confusions that are to be encountered in the philosophy of action, and, I believe, what essen tially prevents a satisfactory solution to the problem of whether or not reasons are causes. Several different and distinct theses about how an action is to be cor rectly described are to be discovered in the literature. The first and by far the most common thesis is the the sis that there is not necessarily one and only one cor rect action-description which applies to an action. This thesis seems to be uncontrovers.ial since everything admits of the possibility of being variously described provided only we have a language which is rich enough to allow for it. A second but distinct thesis is the thesis that a human action can only be said to possess a certain property "under a certain description." Thus, in her book Intention, Miss Anscombe presents us with a case of a man who is moving his arm up and down, pumping poisoned water into a cistern, which supplies the drink ing water of a house which is inhabited by a group of party chiefs, who control the state, who are exterminating the Jews, and planning a world war. Of this case, she writes, Now we ask: What is this man doing? What is the description of his action? First, of course, any description of what is going on, with Him as subject, which is in fact true. E.g., he is earning wages, he is supporting a family, he is wearing away his shoesoles, he is making a distur bance of the air. He is sweating, he is generating those substances in his nerve fibers. If in fact good government, or the Kingdom of Heaven on earth and a good life for everyone, comes about by the la bours of the good men who get into power because the party chiefs die, then he will have been helping to produce this state of affairs. However, our enquiries into the question "Why?" enable us to narrow down our consideration of descriptions of what he is doing to a range covering all and only his intentional actions.132 A closely related thesis is the claim that a human action is capable of receiving a certain sort of explana tion only if it has been described in a certain fashion. Hence, Donald Davidson claims, I flipped the switch because I wanted to turn on the light, and by saying I wanted to turn on the light I explain (give my reason for, rationalize) the flipping. But I do not, by giving this reason, rationalize my alerting of the prowler nor my illuminating of the room. Since reasons may rationalize what someone does when it is described in one way and not when it is described in another, we cannot treat what was done simply as a term in sentences like "My reason for flipping 142G# e . M. Anscombe, Intention (Ithaca, New York: Cornell University Press, 1969), pp. 37-38. 178 the switch was that I wanted to turn on the light1'; otherwise we would be forced to con clude, from the fact that flipping the switch was identical with alerting the prowler, that my reason for alerting the prowler was that I wanted to turn on the light.143 Thus, while some explanatory sentence may succeed in ex plaining what an agent does "under one description" it may utterly fail to explain his actions under some al ternative description. It would appear then that we may succeed in explaining why a particular man is playing a piano by saying that he is practicing for a concert provided we have previously described what he is doing by saying that he is "playing the piano." But we cannot ex plain what he is doing by saying that he is practicing for a concert provided he has already been described as "practicing for a concert." As they stand, these different theses about how human actions may or may not be correctly described are unlikely in themselves to cause any serious difficulties. But serious confusion immediately arises when they are unconsciously identified or combined with a radically different and quite distinct thesis about human actions. The thesis in question here is the thesis that multiple action descriptions are to be thought of as all designating 1 4 3 D o n a l d Davidson, "Actions, Reasons, and Causes," in The Nature of Human Action, ed. by Myles Brand, op. cit., pp. 68-69"! 179 one and the same action. This thesis can perhaps best be seen in terms of the following example: Suppose that John does each of the following things at more or less the same time: (1) he raises his arm, (2) he moves his arm to the right, (3)he frightens a fly on the edge of the desk, (4) he flexes his index finger, (5) he pulls the trigger, (6) he fires the gun, (7) he kills Smith. Now the thesis in question here holds that, with respect to this example, we do not have seven acts performed by John, but what we really have is seven different descrip tions of one and the same act. That, at least, Anscombe and Davidson hold this thesis is evident from their writings. Hence, Miss Ans combe writes: . . . Are we to say that the man who (inten tionally) moves his arm, operates the pump, replenishes the water supply, poisons the inhabitant, is performing four actions? Or only one? . . . In short, the only distinct action of his that is in question is this one, A. For moving his arm up and down with his fingers round the pump handle is, in these circumstances, operating the pump; and in these circumstances, it iss replenishing the house water-supply; and, in these cir cumstances, it is_ poisoning the household. So there is one action with four descrip tions, . . .144 Similarly, Donald Davidson holds the same thesis. 144q . e . M. Anscombe, Intention, op. cit., pp. 45-46. 180 I flip the switch, turn on the light, and il luminate the room. Unbeknownst to me X also alert a prowler to the fact that I am home. Here I do not do four things, but only one, of which four descriptions have been given.145 And elsewhere, But what is the relation between my pointing the gun and pulling the trigger, and my shooting the victim? The natural and, I think, correct answer is that the relation is that of identity.i^6 But if the thesis of the identity of these acts is accepted and conjoined with the thesis that a reason explains an action only under a certain description, then certain irremediable difficulties arise. For it is perfectly possible for a man to intend to move his arm up and down without intending to pump water, replenish the water supply, poison the inhabitants of the house, and so on. But clearly, if the man's intention is to pump water, this may succeed in explaining the action of moving his arm up and down. But it does not explain the action of replenishing the water supply or poisoning the inhabitants since, by hypothesis, the agent had no such relevant intentions and the intention to pump water is not l^^Donald Davidson, "Actions, Reasons, and Causes," in The Mature of Human Action, ed. by Myles Brand, op. cit., p p . 6 8-(T9 . Italics mine. ■I / g Donald Davidson, "The Logical Form of Action Sen tences," in The Logic of Decision and Action, ed. by Nicho las Rescher (Pittsburgh: University of Pittsburgh Press, 1967), p. 84. 181 relevant to the action described as "replenishing the water supply" or the action described as "poisoning the inhabitants." It is precisely this descriptive- relativity of explanations by intentions or reasons that accounts for why so many theorists are unwilling to treat reasons as causes. For, if they are unwilling, as most of them appear to be, to think of causes as description-relative, then having an intention cannot be considered to be a cause of an action. However, on the thesis that there are not four dif ferent acts here, but four different descriptions of one and the same act, it would appear that in giving the man's intention to pump water we would also have an explanation or at least a partial explanation of his acts of replen ishing the water supply and poisoning the inhabitants of the house. And if poisoning the inhabitants and pumping water are the very same acts and the man's having the in tention of pumping water causally explains (in part) his act of pumping the water, then it causally explains (in part) his act of poisoning the inhabitants. Likewise, it follows from the same considerations that if Davidson takes seriously the thesis that a reason may explain an action under one description but not under another, and at the same time wishes to defend his thesis that reason explanations are simply a species of 182 causal explanation, he will be forced either to give up the thesis of the identity of these acts or admit that the man's intention to pump water is not causally relevant to the poisoning of the inhabitants. And since he has explicitly • committed himself to the first three of these theses, it puts him in the embarrassing position of de nying that the man's intention to pump water is not causally relevant to the poisoning of the inhabitants, which is ab surd. Hence, as long as one clings to the thesis that these acts are identical and the thesis that a reason can suc ceed in explaining an act under one description but fail to explain it under another, one cannot avoid the conclu sion that reasons are not causes. In this case, a theorist may either reject the identical act thesis, the thesis that a reason may succeed in explaining an act under one de scription but not under another, or the thesis that reason explanations are a species of causal explanations. And from the fact that the identity of acts thesis is subject to a number of serious criticisms, I chose to defend the others and reject it. Clearly the identity thesis fails first of all be cause these acts which are alleged by it to be identical, do not satisfy the principle of the indiscernibility of identicals, i.e., they fail to have all their properties in common. Alvin Goldman has presented three counter examples in which actions which are alleged to be identical either fail to have identical effects or fail to have identical causes or causal factors. Consider, for ex ample, the act of John's killing Smith and the event of the gun's firing. Now, while it is perfectly legitimate to suppose that John's act of pulling the trigger caused the event of the gun's firing, it would be extremely odd to suppose that John's act of killing Smith caused the event of the gun's firing. Therefore, John's act of killing Smith is not identical with John's act of pull ing the trigger. Likewise, suppose that John performs the act of playing the piano, the act of putting Smith to sleep, and the act of awakening Brown at more or less the same time. Now, although in this case John's act of playing the piano may be said to have caused the events of Smith's falling to sleep and Brown's waking up, John's act of awakening Brown did not cause the event of Smith's falling asleep, nor did John's act of putting Smith to sleep cause the event of Brown's waking up. Therefore, John's act of playing the piano is neither identical with his act of putting Smith to sleep nor with his act of awakening Brown. Finally, consider the situation in which George replaces a burned-out lightbulb in a particu lar socket a moment before John comes along and flips the switch and thereby turns on the light. Now, George's 184 replacing the lightbulb is a cause, or causal factor, of John's act of turning on the light but not of John's act of flipping the switch, since without it John would not have succeeded in turning on the light but he may have flipped the switch without it. And thus, John's act of flipping the switch is not identical with John's act of turning on the light.-*-47 Nor is this the end of the difficulties for the * identity thesis. Goldman also criticizes the view on the grounds that whereas, identity is a symmetrical and re flexive relation, the relationship between acts which is implicit in our use of the "by" locution is both asymmet rical and irreflexive. We would not say, for instance, John flipped the switch by^ turning on the light, but we do say that John turned on the light by^ flipping the switch. Nor would we say that John turned on the light by^ turning on the light, or that he flipped the switch by flipping the switch.148 Another difficulty with the identity thesis is that if it is true, then the distinc tion between basic and non-basic actions will have to be abandoned. For if John's flexing his index finger is a basic action, and it is identical with John's killing Goldman, A Theory of Human Action, op. cit., pp. 2-4. 148Ibid., p. 5. 185 Smith, then John's killing Smith is a basic action.149 These difficulties, to my mind, ,are serious enough to cause any serious-minded philosopher to reject the identity of acts thesis. 4.2. "Doing X by Doing Y" And "Doing X in Doing Y " In everyday language we frequently speak of our selves and others as doing something by doing something else (e.g., we commonly say "I fired the gun by pulling the trigger"), and we also frequently speak of people as doing things iii doing other things (e.g., we commonly say "I issued a check ill signing my name") . Such ways of speaking, or something very much like them, are fundamental to Austin's famous distinction between illocutionary and perlocutionary speech-acts. The various uses to which these phrases are put in everyday discourse has a direct bearing on the question of whether and to what extent an act can be described independently of its antecedents, circumstances, and consequences. Consider first, the use of the "by" locution. The various uses that are made of the "by" locution suggest that one of the ways in which we might look at actions is in terms of actions and the consequences which follow them. For, suppose that John does each of the following things at more or less the 149Ibid., p. 6. 186 same time: (1) he raises his arm, (2) he moves his arm to the right, (3) he flexes his index finger, (4) he pulls the trigger, (5) he fires the gun, and (6) he kills Smith. With regard to this example, it appears as if we are free either to characterize John's action as simply kill ing Smith, or, if we so chose, as firing the gun which then had as a consequence, Smith's death. And we might in this latter case extend the range of the consequences so that they include such things as Smith's family going on welfare, shame being brought on John's entire family, his losing his job, and so on. Joel Feinberg has labelled this the "accordion effect" and describes it as follows: This well-known feature of our language, whereby a man's action can be described as narrowly or broadly as we please, X propose to call the "ac cordion effect," because an act, like the fold ing musical instrument, can be squeezed down to a minimum or else stretched out. He turned the key, he opened the door, he startled Smith, he killed Smith— all of these are things we might say that Jones did with one identical set of bodily movements. . . 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 pur pose. Instead of saying Smith did A (a rela tively 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."150 This supposed accordion effect in conjunction with the fact that the same action can be performed in a vast lSOjoei Feinberg, "Action and Responsibility," op. cit., pp. 106-107. 187 variety of different circumstances may tempt one to claim that an agent performs an infinite number of a c t s . - ^ l That is, it may be argued that since what happens when a person performs an action is capable of being subdivided in an infinite number of different ways into acts, circumstances, and consequences; that there are an infinite number of possible act descriptions for any single act performed by an agent. And consequently, that in performing any single action, an agent is in reality performing an infinite number of actions. Yet such an argument is clearly fallacious. For, in the first place, there is little plausibility in the claim that events and objects can be infinitely subdivided in this fashion. Such a supposition would clearly lead us into the mists of Zeno's paradoxes. Second, the argu ment also mistakenly assumes that multiple action descrip tions are really descriptions of different acts. But although we have already seen that the identity thesis is untenable, there is no good reason to suppose that various action descriptions are not capable of describing the same action. "John's striking the tallest man in the room" may suc ceed in describing the very same action as "John's striking the oldest man in the room." Likewise, "Don’s checkmating ISlgtuart Hampshire comes dangerously close to holding this thesis. C.f. Thought and Action {New York: The Viking Press, 19 60), p. 192. 188 his opponent" may succeed in describing the very same action described by "Don's checkmating Frank Salt." For it is easy to imagine that in these cases the tallest man in the room is identical with the oldest man in the room and that Don's opponent is identical with Frank Salt. In these cases we have what amounts to different descriptions of the same action under the same concept. Furthermore, wherever there is the possibility of in complete description there is also the possibility of equally correct multiple descriptions. These considerations aside, Feinberg's charac terization of the accordion effect is defective. For while it is impossible to draw any sharp line between what will count as part of the description of the action itself and what will count as the description of the consequences of that action, it does not follow that there are no criteria whatsoever for deciding which features of the situation belong to the description of the agent's action and which features belong to the description of the consequences of the agent's action. Some of the features which have been suggested as relevant here are the agent's intentions, temporal remoteness, the general or specific interests of the one making the 189 inquiry or narrating, and the existence of certain prac tices or conventions. ^2 These criteria, however, are at most only very weak criteria and do not apply in each and every case. First, it is not true that a consequence must be intended by the agent in order for it to be elided into the descrip tion of the act. If this were not the case, then it would be logically impossible for someone to act unin tentionally. Second, the temporal remoteness of the consequences is bound to be at most a very vague criterion. For how remote do the consequences have to be before they are too remote? If I set a trap for an animal and the trap does not capture an animal until a year or ten years from now, it is still true that I trapped the animal. Of course, whether or not and precisely when the animal becomes trapped in the trap depends upon many intervening factors which are beyond my control, but how many are too many? Third, the general or spe cific interests of the one making the inquiry or narrating is bound to have an effect on exactly what consequences he will allow to be elided into the description of the action. Thus, if the inquirer's interest lies in morally or legally evaluating whattook place, he might be tempted 152^ncirew oidenquist, "Choosing, Deciding, and Doing," in The Encyclopedia of Philosophy, II, p. 101. to consider John killed Smith as a legitimate description of John's act, while if the inquirer's interest lies in learning how to fire a gun, he would be tempted'to accept only things like John pulled the trigger as the appropriate description of John's act. However, given that interests vary so widely, this criterion cannot be considered as providing us with any real help in distin guishing those consequences which can and those conse quences which cannot be successfully elided into the description of the act. The fourth criterion, viz., the existence of certain practices and conventions, although applicable in some cases, is not applicable in all cases. For although there are some pretty well defined practices or conventions which apply to such things as pointing, signally, checkmating, etc., there are a vast number of actions such as scratching your head, tying your shoelace, raising your arm, etc., for which there are no set of social practices or conven tions. Eric D'Arcy has defended the thesis that not all terms which denote acts may be elided into terms which denote their consequences on the grounds that some acts are of such moral and social significance that this significance would be lost if elision were allowed in 191 their c a s e . 153 The statement of his thesis is as fol lows : Certain kinds of acts are of such signifi cance 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.154 Some examples include killing, maiming, torturing, breaking a contract, betraying, and stealing. But D'Arcy's thesis does not cover a large number of cases, nor does his defense of it provide a conclusive reason why elision,cannot occur even in these cases. For if one of the consequences of my telling a lie is the saving of Mr. Roberts' life, there is no reason why we cannot simply say "I saved Mr. Roberts' life." D'Arcy's thesis seems plausible to most of us simply because the sorts of actions wbich qualify are just the sorts of actions that most of us are interested in preventing. But if interest is made the determining factor here, it must be allowed that any conceivable interest may be used in deciding whether or not elision can take place. In any case, it is clear that whenever elision does occur, the resulting action description is something less than 153j;ric D'Arcy, Human Acts (London, England: Oxford University Press, ,1963), pp. 18-21. 154Ibid., p. 18. 192 a complete action description. Hence, it is always possible to demand a more complete description and hence deny that elision is always a legitimate move. 4.3. Basic Actions Vs. Nonbasic Actions Assuming that human beings perform any actions at all, it follows that they must perform some actions "di rectly." This follows because otherwise we would be caught in the infinite regress of doing Z by doing Y, doing Y by doing X, and so on ad infinitum. This feature has led some philosophers to the view that there are basic and nonbasic actions that human beings perform. Thus, Arthur Danto, in "Basic Actions"155 an(j "what We Can Do"156 has provided definitions of these two kinds of acts. According to Danto, a basic and a nonbasic action may be defined as follows: B is a basic action for a person S if and only if (i) B is an action performed by S, and (ii) there is no other action C such that S performs C and C causes B. N is a nonbasic action for a person S if and only if (i) N is an action performed by S, and (ii) N is not a basic action performed by S. However, Danto's definitions are faulty because together they suggest that a nonbasic action is caused by l55^rthur Danto, "Basic Actions," in The Philosophy of Action, ed. by Alan R. White, op. cit., pp. 43-58. 156Arthur Danto, "What We Can Do," in The Journal of Philosophy, LX (1963), 436. 193 some other action that the agent performed. For consider John's nonbasic act of swatting a fly with a flyswatter. What is the act that John performs which causes him to perform the act of swatting the fly? Certainly, in doing this John performed the basic action of grasping the flyswatter, raising his arm, moving his arm down, and so on. But none of these causes John's act of swat ting the fly. What they caused was the flyswatter to rise and fall. But the rising and falling of the fly swatter are not John's acts but events which John brings about. In other words, Danto's definitions fail to distinguish between the events that one brings about and the acts that one performs. Danto's definitions are also faulty because they would commit us to classifying some basic actions as non-basic actions. To see this suppose that John per forms the act of (accidentally) locking his car door with his car keys in the ignition, and this act, in turn, causes him to break a window with a rock in order to get back in. But in breaking the window with the rock and reaching in to unlock the door John performs many actions which we would ordinarily want to classify 157This criticism, or one very similar to it, has been made by Myles Brand, in "Danto's Basic Actions," Nous, II (1968), .187-190. 194 as basic actions, i.e., he reaches down, .grasps the rock, raises his arm, swings his arm, and so on. Each one of these bodily actions, however, is something which John is caused to do by his act of locking himself out of the car, and, according to Danto's definitions, ought to be classified as nonbasic actions. The former criticism of Danto's definition of a basic action suggests that we must distinguish between the events that a person causes to occur in performing an action and the actions he may or may not perform in or by performing that action. Consider, for example, John's act of'turning on the light, and suppose that John accomplishes this by flipping the light switch. In this case we should distinguish the event that John causes by flipping the switch and the act which he per forms by flipping the switch. The event which John causes is the event of the light's going on, and the act that John performs is the act of turning on the light. The event of the light's going on is what must (logically, not causally) occur in order for John to perform the act of turning on the light. Let us call this event or set of events whose occurrence is necessary to the performance of an act the event counterpart of the act. The notion of an event counterpart makes it possible to redefine the notion of basic and nonbasic 19 5 actions so as to avoid these difficulties. The resulting definition is B is a basic action for a person S if and only if {i) B is an action performed by S, and (ii) there is no other action C such that S performs C and C causes the event counterpart of B. This definition applies to individual actions or what we shall call act-tokens rather than types of acts or what we shall call act-types. On this definition of a basic action bodily actions such as raising one's arm are typical examples of basic human actions. 4.4. Action Types and Action Tokens The possibility of multiple action descriptions is bound to cause some difficulty if we do not dis tinguish between a complete action description, an in dividuating action description, and an adequate action description. A complete action description would be one which specified all the action's details, including all of its relational properties. Thus, to completely de scribe an action we might have to describe everything in the history of the universe. It goes without saying then that any theory of human action must settle for something less than complete action descriptions. An individuating action description is any description which can serve to distinguish that action from other actions. Telling who performed the action, describing what was done in general terms, and pointing are all ways 196 of individuating actions. But such individuating action descriptions, although they may serve to separate that action from other actions, frequently leave out many of the essential descriptive features of the actions. Hence, pointing and uttering the words "That fight" may serve to identify the particular fight in question, but it would also fail to adequately describe what oc curred. An adequate action description, on the other hand, neither enumerates all of the action's detail nor merely identifies the action in question, but serves to indicate the essential or primary aspects of the action. It should be obvious that this contrast between com plete action descriptions, identifying action descrip tions, and adequate action descriptions was not intended to be a distinction between three kinds of mutually exclusive descriptions. Obviously, a complete action de scription would be both an adequate and an identifying action description, and there is no reason why an identi fying action description could not be an adequate action description and vice versa. However, barring the possi bility of ever succeeding in completely describing a human action, the only plausible alternative seems to be to search for action descriptions which are at the same time specific enough to identify particular actions 197 and general enough to allow us to speak of different kinds of actions. .Unfortunately, ordinary language seems to be simply too obscure and indeterminate in this regard. For one thing, ordinary usage frequently fails to dis tinguish between a generic act and a particular act. Hence, in ordinary usage an assassination and the assas sination of John F. Kennedy are both commonly referred to as acts. And, although philosophers have not always explicitly drawn such a distinction, there is some evidence that many of them have nevertheless felt the need of doing so. The need for such a distinction is at least implicit in Aristotle's philosophy, which is evident from his discussion of voluntary action. Hence, he claims, But with regard to the tilings that are done from fear of greater evils or for some noble object (e.g., if a tyrant were to order one to do something base, having one's parents and chil dren in his power, and if one did the action they were to be saved, but otherwise would be put to death), it may be debated whether such actions are involuntary or voluntary. Something of the sort happens also with regard to the throwing of goods overboard in a storm; for in the abstract no one throws goods away volun tarily, but on condition of its securing the safety of himself and his crew any sensible man does so. Such actions, then, are mixed, but are more like voluntary actions; for they are worthy of choice at the time when they are done, and the end of an action is relative to the occasion. Both the terms, then, "vol untary" and "involuntary," must be used with reference to the moment of action . . . Such actions, therefore, are voluntary, but in the abstract perhaps involuntary; for no one would choose any such act in itself. 19 8 And also, But the things that in themselves are involun tary, but now and in return for these gains . are worthy of choice, and whose moving prin ciple is in the agent, are in themselves in voluntary, but now and in return for these gains voluntary. They are more like volun tary acts; for actions are in the class of particulars, and the particular acts here are voluntary. 8 As seems clear from these passages, Aristotle is attempt ing to draw a distinction between actions considered in the abstract (throwing goods overboard), and more specifi cally, particular actions (throwing goods overboard in a storm which was endangering the ship and the safety of everyone aboard). This distinction can also be found more explicitly stated in the works of several contemporary philosophers. Thus, Georg Henrik von Wright in his Norm and Action, dis tinguishes what he calls "generic acts or act-categories" from "individual acts or act-individuals.1,159 Likewise, in the process of arguing that human ability is a species of physical possibility, Storrs McCall claims that we must distinguish between types of actions ("generic" actions), 15®Aristotle, Nicomachean Ethics, in The Basic Works of Aristotle, ed. by Richard McKeon, trans. by W. D. Ross (New York: Random House, 1941), 1110a 4-19, 1110b 2-6. 159 Georg Henrik von Wright, Norm and Action: A Logical Enquiry (New York: Routledge and Kegan Paul, 1963), p. 36. 199 and individual actions.160 The distinction also appears in Nicholas Rescher's "On the Characterization of Ac- tions"-*-61 and "Aspects of Action."162 gut one of the most complete elucidations of this distinction that I know of has been carried out by Alvin Goldman in his book, A Theory of Human Action. And although this distinction is not new with Goldman, his way of drawing it and his system for treating the various relationships between individual acts is at the same time most useful and illuminating. Using the time-honored distinction between types and tokens, Professor Goldman applies these to acts and comes up with the notions of act-types and act-tokens. Since there are many instances of walking, shovelling one's walk, and smoking, these represent act-types. Act-types are individuated into act-tokens by specifying an agent and the particular time at which the act is performed. Hence, John's moving his arm at time t.is a different* act-token than George's moving his arm at time t, and 16°Storrs McCall, "Ability as a Species of Possi bility," in Brand, op. cit., p. 143. ^Nicholas Rescher, "On the Characterization of Actions," Ibid., p. 249. *1 C p ■'■^Nicholas Rescher, "Aspects of Action," in The Logic of Decision and Action, ed. by Nicholas Rescher (Pittsburgh, Pennsylvania: University of Pittsburgh Press, 1966), p. 215. 200 John's moving his arm at time t is a different act-token than John's moving his arm at time t-^, where time t ^ time t-^. Actually Goldman does not offer a definition of act-types. Instead he merely offers several illustra tions, e.g., mowing one's lawn, running, writing a let ter, and giving a lecture; and he adds the general characterization that these are the sorts of things which can be true of or exemplified by a particular subject at some particular time. Act-tokens definitely constitute a subclass of property-instance, but the question of exactly what con stitutes a property and exactly how properties are to be individuated are not only serious problems for Goldman's notions of act-types and act-tokens, they have in one form or another constituted some of the most difficult and serious problems of philosophy in general. In fact, Gold man does not attempt to settle either of these problems completely, but merely commits himself to the adequacy of the synonymy test for identity of properties. On this criterion, properties jzf and 0' are identical if they are expressible by synonymous phrases. This criterion, how ever, is not to be understood as requiring that properties 16^Alvin Goldman, A Theory of Human Action, op.’ cit., p. 10. 201 0 and are in fact expressible by some phrases in some actual language, but only that if^ such phrases were available within some language, then "their being synony mous is a necessary and sufficient condition for their expressing the same property."164 Serious difficulties however seem to attend the synonymy test. First, properties can be identified and designated in a variety of ways. "Brown" and "The color of that dog" both can serve to identify a certain prop erty, they do not succeed in identifying the property in question in the same way. In general, definite descrip tions like "The color of that dog" merely refer indirectly or incompletely to properties. The most serious diffi culty with the synonymy test, however, seems to be with regard to relational properties. Consider, for example, the relational property marked out by the act of striking the shortest man in the room. Now, if, in fact, the shortest man in the room is also the richest man in the room, the property of striking the shortest man in the room might be thought to be identical with the property of striking the richest man in the room. In light of this difficulty, Goldman claims that some revision in the synonymy test may be in order, although he himself decides that he will stay with the test and accept the consequences l64ibid., P. 12. 202 of this for the individuation of acts. This move, I believe, is perfectly justified even though the question of synonymy has occasioned its fair share of controversy in recent years. The resolution of many serious philo sophical problems depends in large part on discovering a precise definition of synonym, but one cannot solve all problems simultaneously. In any case, it would be foolish to demand that philosophers, psychologists, or social scientists formulate a precise definition for every act-property like turning on the light, or check mating your opponent. There may in fact be an indefi nitely large class of different bodily movements which generate the act of breaking a promise and hence no laws to the effect that whenever X, Y, and Z, then a person breaks his promise, but this is no reason to sup pose that these act-tokens cannot be recognized as cases of breaking a promise nor any reason for supposing that not every act-token is caused. The distinction between act-types and act-tokens affords us a conceptual tool for removing much of the confusion which has been a part of many philosophical discussions concerning human actions. Individuating various different act-tokens and act-types in this fashion makes it possible to determine ’.and describe the various relationships which exist between them in a much more . 203 illuminating way than is usually the case in ordinary discourse. Likewise, it releases us from the cumbersome and for the most part unilluminating practice of talking about acts "under certain descriptions, "165 an(j talk about one act being another act "under certain circum- stances."AO The essential structural relationships between various acts can be more clearly described in terms of the notion of level or act-generation. 4.5. Act Generation The ideas which are for the most part only implicit in our everyday uses of the phrases "S did . . . by doing ...," and "S did . . . in doing ___ ," have unfor tunately been the source of many confusions in the philoso phy of action. Fortunately, Alvin Goldman has introduced a notion which promises to be very useful in removing a large part of that confusion. Goldman introduces four categories of what he calls "level-generation" which, although they are not intended by him to be mutually ex clusive, are nevertheless claimed by him to be exhaustive. His four categories of level-generation which allow us to move from one action description to another are: 165C. Donald Davidson, "Actions, Reasons, and Causes," op. cit. A. I. Melden, Free Action, op. cit., pp. 91, 99, 138-139, 142-143, 210. 204 (1) Causal generation, (2) Conventional generation, (3) Simple generation, and (4) Augmentation generation. Causal generation is probably the most frequent form of level-generation between acts. For whenever a person performs an act A which has a certain effect E, he may also be said to perform another act A*. Thus, John's act of moving his hand may causally generate John's act of frightening a fly away, and John's act of checkmating his opponent may causally generate his act of giving his opponent a heart attack. In general, act-token A causally generates act-token A 1 only if A causes E, and A' consists in the agent's causing E. Act-token A con ventionally generates act-token A ’ only if act-token A in circumstances C (possibly null), plus a rule R which says that A done in circumstances C counts as another act-token A'. Hence, John's moving his queen to king- knight-seven may conventionally generate John's check mating his opponent. Simple generation is exactly like conventional generation except without the rule or rules. That is, in simple generation the existence of certain circumstances plus a certain act-token A generates act- token A'. Thus, for example, John's asserting that p may, depending on the circumstances, generate the act- token of breaking his promise, the act-token of contra dicting one of his earlier statements, or the act-token 205 of lying, and so on. One act-token augmentationally generates another act-token whenever the first is per formed in a particular manner or in certain circumstances. Hence, the act-token of John's saying "hello" may aug mentationally generate the act-token of John's saying "hello" loudly, or John's running may augmentationally generate the act-token of John's running at eight m.p.h. In general,•act-token A augmentationally generates act- token A' when A is performed in a certain manner or in certain circumstances. 16*7 These four forms of act generation which may be interpreted as constituting transformation rules for moving from one act description to another, have some important and undeniable advantages over many of the other attempts at fashioning a logic to deal with actions. For one thing, the notion of act generation is not itself dependent on the notion that something is not an act unless it was performed intentionally. For we want, in talking about the acts that agents perform, to be able to speak not only of intentional acts but unintentional may appear as though augmentation generation is merely a species of simple generation since all that is required in either case in addition to the generating act is some set of circumstances. Indeed, as Goldman himself admits, the only difference between them is that in agumentation generation, but not in simple generation, the performance of the generated act entails the perform ance of the generating act. 206 acts as well. Attempts at analyzing the notion of act- generation in terms of appropriate purposes, intentions, and/or goals of the agent fail for the simple reason that many acts are generated by other acts where there are none of the required purposes, intentions, or goals present. Thus, for instance, my act of foreclosing on John's mortgage may cause him to file for bankruptcy, although I did not do it in order to bankrupt him, nor was it my intention to do so. Likewise, I may send a bouquet of flowers to a young lady which consequently results in her suffering an attack of hay fever, but it would not be correct to assert that I necessarily did it for the reason of causing her to suffer an attack of hay fever. The second advantage that the notion of act generation has over other attempts at analyzing acts is that it does not depend on the assumption that acts may be reduced without remainder to events or sets of events. Thus, the notion of act generation enjoys an advantage over various attempts to analyze action sentences in terms of events.1*>8 Georg Henrik von Wright has attempted to develop a systematic 168c.f. Georg Henrik von Wright,' Norm and Action: A Logical Enquiry (New York: The Humanities Press, 1963); and Donald Davidson, "The Logical Form of Action.Sentences," in The Logic of Decision and Action, ed. by Nicholas Rescher (Pittsburgh, Pennsylvania: University of Pittsburgh Press, 1967) . logic of action as an extension of the logic of events, which is in turn an extension of propositional logic. The rationale for this is that actions can only be per formed provided that events occur, and events are the transition between various states of affairs which can be described by various propositions. In von Wright's system "unchange" as well as a change is counted as an event, so that for each state of affairs, one of the fol lowing four elementary change descriptions may apply: pTp, pT-p, —pTp, or -pT-p; and for the state description p & g there would be the four following elementary change descriptions: (pTp) & (gTqJ, (pTp) & (gT-q), (pT-p) & (qTq), and (pT-p) & (qT-q). (The T is von Wright's symbol for the transition operator and reads ". . . is a transfor mation (transition) to - - In von Wright's system then, the elementary act which is the result of an ele mentary change is symbolized by the prefix d. Thus, if p represents the statement that the window is closed, d(pT-p), represents the elementary generic act of the win dow's being opened. But von Wright's d-operator does not by itself ex plain why a particular transition from one state of af fairs to another state of affairs should be considered as an action. Nor does it allow us to move from one act 208 description to another in the way that the notion of act- generation does. The assumption that actions are performed only if events occur, or that every action has an event counter part which must occur if the action occurs, has some dire consequences for the thesis that actions are not causally determined, but free. For clearly, John's checkmating his opponent, John’s signalling, and John's flipping the light switch are all cases in which some form of bodily action and hence, some set of bodily movements, must occur. There is no way in which John can checkmate his opponent, signal, or flip a light switch without performing some bodily action and there is no way in which John can per form a bodily action without some bodily movement occurring. And even if we follow Melden in claiming that these bodily actions are not caused, the necessary event counterparts may be and obviously are caused. Thus, it is difficult to see how human freedom can be defended by defending the thesis that at least human actions are not caused. In much the same vein it is also difficult to see how a theorist can claim that human actions are uncaused and at the same time speak of them as causes. For if they are the sorts of things which can be causes, then they are the sorts of things which can themselves be caused. Nor 209 will it do to claim here, as has been done,*^^ that really what many contemporary action theorists have been arguing is not that actions are uncaused, but that they, like num bers, possibilities, and jokes, are not the sorts of things of which it makes sense to ascribe or deny a cause to. For if they are just not the sorts of things of which it makes no sense either to ascribe or deny a cause to, then they are just the sorts of things of which it makes no sense either to affirm or deny are themselves causes. To claim that actions cannot be causes, is, however, contrary to the way in which we ordinarily speak about actions. Thus, we have absolutely no reservations about saying such things as John's flipping the switch caused the light to go on, or John's infidelity caused his wife to divorce him. We should be on our guard against confusing causation with causal generation. The failure to make this distinction was, as we saw, the most serious defect in Danto's defi nitions of basic and nonbasic acts. Thus, for instance, S may lock himself out of the house, and this may cause him to break a window in order to get back in. The act of locking himself out of the house causes, but does not causally generate, his act of breaking the window in order to get back in. The reason for this, Goldman explains, is ^■^^Raziel Abelson, "Doing, Causing, and Causing to Do,", in The Journal of Philosophy, LXVI (1969), p. 178. 210 Level-generation is precluded between these acts because one of them is subsequent to the other. In cases of causal generation, a causal relation obtains between act-token A and event E, not between act-token A and act-token A'. S's act of flipping the switch causes the event of the light's going on. And since S's turning on the light consistsTn S's causing the light's going on, we may say that S's flipping the switch.causally- ■ -* generates J— S's turning . 3 on the light.17D The temporal requirements on level-generation between act-tokens which do not apply in the cases in which one event causes another, can be more clearly understood if we reflect for a moment on the temporal requirements of the notion of level-generation between property instances. Goldman's definition of level-generation among property- instances is as follows: Property-instance A level-generates property- instance A' if and only if (1) A and A' are distinct property-instances of the same subject; (2) the properties of which A and A' are instances do not differ merely in containing different individual concepts of the same object; (3) neither property-instance A nor property- instance A' is subsequent to the other; nei ther is a temporal part of the other; and they are not co-temporal; (4) there is a set of conditions C* such that 1^°Alvin I. Goldman, A Theory of Human Action (Engle wood Cliffs, New Jersey: Prentice-Hall, Inc., 1970), p. 23. 211 (a) the conjunction of A and C* entails A', but neither A nor C* alone entails A'; (b) if the subject, A, had not exemplified property A (at t), then S would not have exemplified property A' (at t) ; (c) if C* had not obtained, then even though S exemplified property A (at t), S would not have exemplified property A' (at t) Conditions (4)(b) and (c) reflect the asymmetrical character of level-generation. Thus, John's moving his queen to king-knight-seven may conventionally- generate John's checkmating his opponent but John's checkmating his opponent does not generate John's moving his queen to king-knight-seven. Condition (1) clearly marks out the irreflexive character of level-generation. We may succeed in explaining how John checkmated his opponent by saying that he moved his queen to king-knight-seven, but we can not explain how he did this by saying that he checkmated his opponent. Condition 3 states the relevant temporal requirements of level-generation. The non-subsequence requirement' is necessary since it would be incorrect to say that John moved his queen to king-knight-seven "and then" checkmated his opponent, and also incorrect to say that John flipped the switch "and then" turned on the light even if the light does not go on until a few seconds after the switch is closed. But if we accept the notion that 171ibid., p. 45. 212 causes must precede their effects, then it is obvious why causation cannot hold between act-tokens. The requirement that neither the generated act nor the generating act is a proper temporal part of the other is necessary since many acts are composed of a series of smaller acts which are temporally ordered. Hence, typing the word "rose" consists of typing the letter "r," followed by typing the letter "o," followed by typing the letter "s," and so on. And finally, the non-co-temporal requirement is necessary to rule out those acts which are simultaneously performed by an agent but nevertheless constitute com pletely independent acts such as scratching your head as you are wiggling your toes. This difficulty, which is involved in attempting to formulate precise criteria for individuating act-types, is reflected in the important role that the notion of "cir cumstances" plays in each of the four kinds of act- • generation.^72 reason for this is simply that, given 172 It will be recalled that in Goldman's formulation of conventional generation it is possible that the circum stances C are "possibly null." This I believe is a mistake. Goldman explains that the circumstances may be null here be cause, for example, the rule forbidding promise-breaking plus the fact that S broke his promise are in themselves sufficient to warrant our ascribing to S the act of doing what he ought not to do. But this account depends upon con struing such rules in such a way that they do not allow for exceptions, and the question of whether there are any such rules as opposed to mere summary rules is certainly a question which is open to debate. 213 the ways in which we ordinarily classify something as a particular kind of action, there are an indefinite number of circumstances or sets of circumstances in which that kind of act can be performed. .Review, .for example, the numerous ways in which John can perform the act of outjumping George. The relevant circumstances in each case will be the distance that George has just jumped which can encompass any distance short of some undetermined minimum or maximum distance. The impossibility of speci fying exactly what set of circumstances is involved in the performance of various action types is a feature of ordinary everyday uses we make of the "by" and "in" locutions and other action locutions. In this sense, then, it is fruitless to search for some one thing which is, in each and every set of circumstances, a human action. CHAPTER V A CAUSAL ACCOUNT OF HUMAN ACTION In Chapter IV, I argued that some philosophers, Donald Davidson in particular, are committed to a number of incompatible theses concerning human actions and how they are to be described. These three theses are (1) the thesis that some particular set of act descriptions all individual ly describe one identical action; (2) the thesis that a reason may succeed in explaining an action under one descrip tion but fail to explain it under another description, and (3) the thesis that reason explanations are a species of causal explanations. And since it was argued that of these three only one, viz., the first, could be shown to be false while the others seemed quite plausible, it followed that some more precise way of individuating acts had to be discovered. In hopes of discovering such criteria, the question to what extent an action can be described independently of its ante cedents, circumstances, and consequences was discussed. And with regard to the question of to what extent it is pos sible to separate out the act from its consequences it was found that it is impossible to draw any sharp line between what will count as part of the description of the act itself 214 and what will count as part of the description of the con sequences of that act. There were, we found, some very weak criteria for doing this, but they all turned out to be neither necessary nor sufficient in every case. Next, it was argued that not all human actions could be substi tuted for X in the formula "S did X by doing Y," and hence, at least some human actions were basic. It was found, how ever, that Arthur Danto's definitions of a basic and a non- basic action were faulty, and new definitions were suggested which avoid these difficulties. And finally, it was claimed that Alvin Goldman’s act-token and act-type distinction would provide us with the sought-after technique for distinguishing various different individual acts, and that his notion of act-generation went a long way toward making explicit the structural relationships between acts which is only implicit in our everyday use of the "by" and "in" locutions. In this * chapter, I want to argue that a completely adequate causal account of human actions and human ability is possible. 5.1. Intentional Acts In Intention,173 Miss G.E.M. Anscombe has attempted to develop a theory of intentional action in which an intentional 173q .e .M. Anscombe, Intention, (Ithaca, New York: Cor nell University Press, 1969). 216 action is viewed as something which is an indivisible whole. That is, on Anscombe's view intentional actions may not be divided into bodily movements plus an intention. Apparently, she views the term "intentional" as having reference to a form of description of events, a form of description to which a certain sort of "Why" question is applicable. Events are typically described as intentional when the phrase "In order to" or "because" (in one sense) is attached to their descriptions, e.g., I skipped because I felt cheerful.174 Hence, according to Anscombe, one can succeed in giving a different significance to an event such as a bodily move ment by describing it in various different ways, and it is this supposed significance of movements which makes them actions. And, if I understand her correctly, this signifi cance is attachable to the bodily movement provided that (1) the agent was aware of what he was doing at the time he was doing it, (2) the agent was not aware of this through any process of observation, and (3) there is at least the possibility of characterizing what the agent did as the result of mental causality.I?5 174Ibid., pp. 84-85. •*~7^Ibid., p. 25. A similar view is to be found in Stuart Hampshire, Thought and Action (New York: The Viking Press, Inc., 1959), p. 74 ff . 217 Anscombe's account, however, suffers from some serious defects. The primary difficulty with her view centers around the second claim that the agent himself is aware of what he is doing but not through any process of observation. Stuart Hampshire 176 and H.L. Hart 177 have also argued that there is a kind of non-inductive knowledge, or perhaps even certainty, about one's own future actions. The claim that I can. know what I am going to do in the future non-inductively is certainly just plainly false. The first reason it is false is that there is no way in which I can know, except by induction, what I may be able to do in the future. It is an undeniable fact that human beings, like most other things in the universe, lose as well as gain skills and abilities, and the only means of determining whether or not I will retain a particular skill or ability in the future is on the basis of my past, performances. Thus, for instance, the only war rant I have for believing that I will be able to type a passage at the typewriter is the fact that I have performed similar typing tasks in the past. The same applies equally to my future ability to perform even the simplest tasks such as moving my arm. 176 stuart Hampshire, Freedom of the Individual, (London, England: Chatto and Wendus Ltd., 1965). 177 H.L. Hart and A.M. Honore, Causation in the Law (Ox ford: Oxford University Press, 1957). It may be claimed here that what these theorists have in mind is not that we have any. '’direct" non-inductive know ledge of what we will do in the future, but rather, since we know our own reasons for acting without observation and in duction, we can, in this sense, know what we will do in the future. This seems like a more satisfactory claim, but, as I have argued in Chapter III, the claim that we know our own reasons for acting without observation or induction or at least some form of inference is both extremely obscure and surely unjustified. Nor even if we grant that there is some non-observational and non-inferential way in which a person knows his own reasons, it does not follow that he knows what he will do in the future. It does not follow for the simple reason that a person cannot know non-inductively what he will want or intend to do in the future. A person may know now what he wants or intends to do in the future, but that is not to say the he either knows what he will do or knows now what, in the future, he will want or intend to do. It might be thought that we have been interpreting the above view in too strict a fashion. That is, it might be thought that what is being claimed is not that a person can know in some non-observational and non-inferential way what he is doing at the moment he is doing it. But even that this weaker claim is false can be readily seen from an example. Suppose that I am playing the children's game of pinning the 219 tail on the donkey. Blindfolded, there is no way of telling whether I have performed the act of successfully pinning the tail on the donkey or not, although I can perhaps tell that I am moving my legs and raising my arm, but not that I am hitting the target. In her book, Anscombe considers a supposed case in which a statement of an agent's intention is falsified by what he does. And she claims that in such cases the mistake is one of performance rather than one of judgment. Hence, she writes: But is there not possible another case in which a man is simply not doing what he says? As when I say to myself "Now I press Button A" - pres sing Button B - a thing which can certainly happen. This I will call the direct falsification of what I say. And here, . . . the mistake is not one of judgment but of performance. That is, we do not say: What you said was a mistake, because it was supposed to describe what you did and did not describe it, but: What you did was a mistake, because it was not in accordance with what you said. It is precisely analogous to obeying an order wrong - and we ought to be struck by the fact that there is such a thing, and that it is not the same as ignoring, disregarding, or disobeying an order. If the order is given "Left turnl" and the man turns right, there can be clear signs that this was not an act of disobedience. But there is a discrepancy between the language and that of which the language is a description. But the discrepancy does not impute a fault to the lan guage - but to the event. ^78 178 G.E.M. Anscombe, Intention, op. cit., p. 57. 220 But why is it that in this case we must say that it is not the statement of intention which is not mistaken, but rather his performance? Anscombe obviously just assumes that we take the statement "Now I press button A" as a truthful statement of intention. That is, we simply do not question the truth of his statement of what he intends or means to do. This may indeed be a correct observation about what we commonly do, but it is by no means clear that we are justified in doing so. In fact, I believe that there are no conditions or sets of conditions which are both necessary and sufficient for the determining of the truth of intention . statements. To see this let us begin with the supposed fact that we ourselves know when we either intend or do not intend to do something. How do we ourselves discover that we intend to do some thing? This question may strike us as extremely foreign and perplexing, especially if we presuppose that an intention is some sort of mental cause. For, given this interpretation, the question sounds strange precisely because we would like to say that we do not in any sense of the word "discover" that we intend to do something. To discover something im plies that one is looking for something which is somewhat hidden, i.e;, something which is not immediately apparent. But since we suppose that we are always immediately con scious of everything that goes on "in our own minds," we 221 are inclined to say that the correct answer to the question, "How do you know that you intended to go to the store for a pack of cigarettes?" is simply, "Because I know." However, not wishing to sound hopelessly perverse, I think that we are entitled to question these assumptions. To do so let us look closely at an avowal of an intention. Suppose Mr. Staple says, "I intend to pay my light bill this Saturday." Is it enough that he merely says this for us to conclude that he does in fact intend to do it? Now I can quite imagine that there are some who are perfectly content to let the question lie right there. That is, there are those who might be perfectly satisfied to claim that when ever anyone says that he intends or does not intend to do something that is the end of the matter, and there is ab solutely no way of determining whether he did or did not intend to do what he said since, in the end, only he knows. This seems to me to be one of the logical consequences of the mental cause theory, and although I have no intention of denying that such things exist, I do not see how they could provide us with a satisfactory solution to the problem presently before us. In the end such private mental events seem to represent a retreat to indefinables, and it seems obvious that Wittgenstein was quite correct when he claimed that in the case of such indefinables, nothing would do as well as something. 222 Avowals of intention or the lack of such cannot be taken as a sufficient condition for the presence or absence of an intention. Mr. Staple does not intend to pay his light bill this Saturday if, for instance, he makes for the near est new-car dealer the moment he receives his pay check. Nor does he intend to pay his bill if he lies around the house all day Saturday until the final hour for paying his bill has come and gone, in spite of his wife's continued in sistence that he pay the bill. Avowals of intention are not therefore sufficient, and are dismissed if the agent's ac Another additional feature besides the actual perform ance or attempted performance of the act which we frequently take into account in assigning intentions or the lack of intentions is premeditation. We commonly are more inclined to assert that a person really intended to do something if there is evidence of a process of planning that he.went through before he performed the said action or failed to perform it. And the more elaborate the planning was, the more inclined we are to say that he intended to do it. For example, if, in March, a man claims that he intends to visit Europe in April, but he makes absolutely no attempt to ar range for the necessary shots and passport, or to buy plane or boat tickets, or make the necessary hotel reservations, then we are very much less inclined to say that he intends 223 to go. Therefore, antecedent planning is at least one of the criteria we use in assigning or refusing to assign in tentions to agents, but it is hardly a necessary condition. All of this, of course, assumes that the agent has some set of relevant beliefs and desires. For it is by no means inconceivable that a man does really intend to visit Europe in April but he makes no attempt to arrange for the necessary shots and passport, or to buy plane or boat tickets, or to make the necessary hotel reservations if he does not believe that any of this is necessary and/or he has no aversion to the consequences of not making such pre liminary preparations. If, for instance, he does not know or believe that a passport is necessary, he will make no attempt to obtain one although it may still be true that he intends to visit Europe in April. Or he may not have any aversion to sleeping on park benches or at the side of the road, and hence, make no hotel reservations. Thus, the mere lack of planning is not a sufficient condition for asserting that an agent did not intend to do something. Nor, would it appear, that the presence of planning is either a necessary or a sufficient condition for the; presence of an intention. It is not a necessary condition unless we are prepared to say that nothing done "on the spur of the moment" could be an intentional act. If a man immediately 224 strikes his neighbor for making an insulting remark about his mother rather than going several miles and fetching a weapon with which to wound him, we are still inclined to say that he did in intentionally. It should be noted here that frequently the length of time which expires between the occurrence of the insult and the act is an important factor in our ascription of intent. Generally the longer the interval, the more we are inclined to attribute an intention to the agent although it is doubtful whether we can draw any sharp line here. The fact that we are not able to draw any precise line is merely a reflection of the fact that as we ordinarily use the word "intention" it applies to anything a person does as long as he can be said to have some sort of a dim awareness of what he is aiming at the time he is doing it. The presence of planning is not a sufficient condition because certain activities are frequently ambiguous as to their intended result. Usually a man who sits at a type writer writing away for months, turning out page after page, is not doing it just to pass the time away, but it does not follow that he intends to publish it, especially if he has plenty of opportunities to do so but does not. Therefore, neither the presence nor absence of planning can be taken, by itself, as either a necessary or a sufficient condition for ascribing or refusing to ascribe an intention to an 225 agent, and Anscombe's claim that the mistake lies in the performance rather than the statement of intention is not justified. Nor is it true that Anscombe's claim that there exists a precise analogy between this case of announcing that you are about to push button A and then push button B, and the case of obeying an order wrongly, justified. For, if what I have been arguing is correct, there are no absolutely clear signs that when a man turns right when given the order "left turn," he is not disregarding or disobeying the order. The analogy is not as precise as Anscombe thinks it is for other considerations as well. For if what many philosophers claim is the case is true, then orders and commands are not reports of events, mental or otherwise. And if this is the case, then the only sense in which we may plausibly speak of the "falsification" of an order or command is the sense in which what was done fails to accord with what was ordered done. But if, on the other hand, avowals of intention are "falsifiable" in the sense of failing to be correct "reports" of what the agent actually intends to do, then they are hardly perfectly analogous to orders or commands. Thus, Anscombe's analogy and the claim that there are two knowledges - one by observation, the other in intention - which she bases on the analogy, are hardly enough to be convincing. In view of Anscombe's failure to account for the fact that we do need observation to determine whether or not we have succeeded in performing some act which is a consequence or result of some more basic bodily act, it may be argued that at least we can tell what bodily actions we are per forming at the time we are performing them without the aid of observation or induction. But although such a claim may be made, it is by no means clear that it can be justified. For it is not clear that the notion of observation must be narrowly restricted to whatever is found out by the five senses. There is no reason why I cannot find out that I have moved my muscles and thus moved my arm, by means of kinesthetic sensations. Clearly, I can determine that I moved my arm, although I am prevented from seeing it move by feeling it move, and whatever is found out by means of such kinesthetic sensations is not radically different from what is found out by means of the five senses. 5.2. The Justification of a Causal Account of Intentional Action The distinguishing characteristic of purposive action is that it is the kind of behavior which the agent does for a reason, and, as we have seen, many action theorists (Melden, Anscombe, et al.) have attempted to explicate the difference between a human action and a mere bodily move ment in terms of the presence of absence of an appropriate 227 reason. Unfortunately, the exact sense in which reasons account for this difference has, in spite of all of their efforts to explain it, remained obscure. However, I be lieve that the exact sense in which reasons account for this distinction can be explicated by conceiving of them as causes of basic bodily movements. What is it then to do something for a reason? Suppose that John's reason for flipping the switch is to turn on the light. John flipped the switch then in order to turn on the light. But what is the meaning of this "in order to”? How does John's turning the light on explain John's flipping the switch? The answer, I believe, is that it does not. John's turning on the light is obviously not a sufficient condition for the truth of the sentence "John flipped the switch in order to turn on the light." For the mere fact that John turned on the light does not tell us that John flipped the switch in order to turn on the light, since the mere fact that John turned on the light by flipping the switch does not imply that he may not have had some other reason for flipping the switch. He may, for instance, being unacquainted with electrical switches, have flipped the switch merely in order to find out what would happen. He may have flipped the switch in order to make a noise which would, he hoped, attract someone's attention, or he may have flipped the switch in order to activate the garbage disposal. 228 But not only is the fact that John turned on the light not imply that he flipped the switch in order to turn on the light, it isn't even relevant to the question of whether John flipped the switch in order to turn on the light. It isn't relevant simply because John may not succeed in turn ing the light on (imagine no lightbulb in the socket) even though it is still true that John flipped the switch in order to turn the light on. What is implied by "John flipped the switch in order to turn on the light" is neither that he succeeded in turning on the light, nor that he actually turned on the light by flipping the switch, but rather, that he wanted to turn on the light and he believed that by flipping the switch he would succeed in turning on the light. But John's wanting to turn on the light and his believing that by flipping the switch he would turn on the light are not themselves suf ficient to explain his act of flipping the switch, since these are consistent with John's flipping the switch for some other reason, or for no reason at all. That is, John may have some additional (and perhaps stronger) interest or desire for flipping the switch as when his primary interest in doing it is not to turn on the light, but to make a noise so as to annoy his parent. Likewise, John may flip the switch for no reason at all (accidentally) even 229 though he wants to turn the light on and he believes by flipping the switch he will accomplish this task, but while in the process of reaching for the switch he notices a glass about to fall off a shelf near the light switch and, while in the process of reaching for the glass, his arm accidentally flips the switch. Clearly, in this case, it is false that John flipped the switch in order to turn on the light, while it is true that he both wanted to turn the light on and believed that flipping the switch would do it. What else is needed in order to account for the truth of "John flipped the switch in order to turn the light on."? I believe the most plausible and straightforward answer is that John’s wanting to turn on the light and his believing that by flipping the switch he would, caused him to flip the switch. If John's want and belief do not cause him to flip the switch in this way, then John's flipping the switch is not intentional but accidental. It does not follow from this that all acts are inten tional. Some philosophers have been mistakenly led to the view that all acts must be intentional from the fact that some acts or act descriptions imply or presuppose certain specific wants and beliefs on the part of the agent. Thus, to describe John as hiding from Frank because he has aligned his body behind a tree in such a way that no part is visible 230 to Frank presupposes or implies that he is doing so out of a desire not to be seen by Frank. Likewise, to describe some one as fishing implies or presupposes that he is dangling a line with a baited hook in the water for the purpose of catching fish. But although many acts and act descriptions thus presuppose or imply certain wants and beliefs on the part of the agent, it is not true that all acts or act de scriptions presuppose or imply certain specific wants and beliefs on the part of the agent who performs them. For there is a large class of acts which can be done accidentally. These include turning the light on, hitting another car with your own, hitting your thumb with a hammer instead of hitting the nail head, and so on. 5.3. Goldman's Account of Intentional Action If what I have been arguing is correct, i.e., that intentional actions, at least, are caused by some set of the agent’s wants and beliefs, then it would appear that some sort of a causal account would hold the greatest promise. In fact, Alvin Goldman has presented just such a causal an alysis in his A Theory of Human Action which is, I believe, the most complete and well thought-out analysis I have encountered. For in presenting his analysis of an inten tional action, Goldman brings in some notions which are quite useful in solving many of the other problems in the philoso phy of action as. well. 231 Goldman's analysis of an intentional action proceeds in terms of three other concepts: (1) the concept of act- generation? (2) the concept of a basic act-token; and (3) the concept of an action-plan. The reader is already familiar with the concept of act-generation from Chapter IV, and this concept, together with the concept of a basic act- token, is, I believe, sufficient to account for one of the major contrasts philosophers have wanted to make in talking about human actions, i.e., the contrast between my moving my arm and my signalling. Goldman's definition of an intentional act is as follows: Suppose S has an action-plan which includes acts Ai, A2f A3, . . . An , where A1 is a basic act and n >_ 1. S wants to do An , and S>believes (to some degree) of each of the acts A^, A2, A3 , . . . An firstly, that it will either be generated by A^ or be on the same level as A^, and secondly, that it will either generate An or be on the same level as An . if this action-plan, in a certain character istic way, causes S's doing Aj_, then A^ is intentional. And if some of the other acts A^, A3 , . . . An are performed in the way conceived in the action-plan, then these acts are also intentional. All other acts on the (actual) act-tree are non-intentional.-179 This definition appeals to the notions of a basic act-token and an action plan. Goldman defines a basic act-type as follows: Property A is a basic act-type for S at t if and only if 179 Alvin Goldman, A Theory of Human Action, op. cit., p. 57. 232 (a) If S v/ere in standard conditions with respect to A at t, then if S wanted to exemplify A at t, S's exemplifying A at t would result from this want; and (b) the fact expressed by (a) does not depend on S's level-generational knowledge nor on S's cause-and-effeet knowledge, except possibly the knowledge that his exempli fying A would be caused by his want.180 Condition (a) however, contains the as yet undefined notion of a basic act-token. Goldman defines this as follows: S's exemplifying property A (at t) is a basic act-token if and only if (a) property A is a basic act-type for S (at t); (b) S's exemplifying A (at t) is caused, in the characteristic way, by an action-plan of S; and (c) S's exemplifying A (at t) is not level generated by anything else satisfying clauses (a) and (b), except possibly by augmentation generation.181 All of these definitions stand in need of further elaboration. Let us begin with clause (a) of the definition of a basic act- type. 180ibid., p. 67. 181ibid., p. 72. 233 The first thing that needs to be noted about part (a) of the definition is that it is a complex hypothetical with a consequence clause which is itself a subjunctive proposition. The subjunctive, viz., "if S wanted to exemplify A at t, S's exemplifying A at t would result from this want," is necessary, Goldman explains, because the essential feature of a basic act-type or basic-act property is that it is the kind of property that can be exemplified by at least some agents at will.3*82 This has the effect of ruling out such things as sneezing, hiccuping, salivating, vomiting, blushing, and fainting as instances of basic act- types, and since every basic act-token is an instance of a basic act-type, it rules them out as basic act-tokens as well. This subjunctive conditional is Goldman's interpreta tion of the sentence "S can (cannot) exemplify property A at will." And although he uses the word "result," Goldman wants this subjunctive conditional to be interpreted as a causal subjunctive which refers to a dispositional property of the agent. In another place, he claims that this causal condition statement, viz., "If S wanted to do A at t, then S would do A at t," is a sufficient condition for or entails 18 2it is the task of the behavioral sciences, Goldman tells us, to discover the precise range of our basic action- types. See p. 89. 234 the truth of the statement "S is able to do A at We shall have an opportunity to examine and defend Goldman's definition of ability in more detail in what follows. It may be thought that the remaining part of condition (a)/ v i z ., "If S were in standard conditions with respect to A at t,l! lends a note of obscurity to the notion of a basic act-type. The necessity of including such a notion as standard conditions is brought about by the fact that fre quently an agent cannot perform a basic act because some thing external inhibits, for example, although it may be the case that an agent can, in ordinary conditions, simply raise his arm, if his arm is tied down, he cannot. Likewise, it is not possible for a man to raise his arm even if that action is in his repertoire of basic actions if his arm is already up. In general then, the notion of standard condi tions can be characterized as follows: S is in standard conditions with respect to property A at t just in case (a) there are no external forces at t making it physically im possible for S to exemplify A at t, and (b) if exemplifying A involves a change into state Z, then S is not already in Z.184 183ibid., pp. 199-200. 184Ibid., p. 65. 235 It is curious to note here that Goldman does not include in the notion of standard conditions the internal states of S ’s body. But the decision to rule out internal states as irrelevant is unjustified, and, at best, merely arbitrary. The remaining part of Goldman's definition of a basic act-type, viz., S's exemplifying A at t would result from this want, clearly indicates that reasons are to be treated as causes. Actually, to attribute a basic act-type to an individual is simply to attribute a certain kind of disposi tional property to him. Hence, to assert that an individual possesses the basic act-type of raising his right arm at time t is to assert that he has the dispositional property of exemplifying the act of raising his arm as a result of his wanting to raise it. All of this assumes that wanting to raise his arm is the only want, or at least the only relevant want, the agent had.And, as we shall shortly see, certain complications arise when the agent has stronger or equally strong conflicting desires. The notion of level or act-generation included in part (b) of Goldman's definition of a basic act-type, viz., the fact expressed by (a) does not depend on S's level-genera- tional knowledge nor on S's cause-and-effeet knowledge, 185jbid., p. 86. 236 except possibly the knowledge that his exemplifying A would be caused by his want, has already been explained. The re maining section, viz., (a) does not depend on S's cause-and- effect knowledge, is necessary to avoid certain cases in which the agent himself can by performing a basic action cause his body to react in a certain way. Thus, almost every one can cause themselves to vomit by merely putting their fingers down their throats. But this is a clear case of the employment of the agent's cause-and-effeet knowledge, since he is able to bring this about only because he knows that by sticking his finger down his throat he can cause himself to vomit. Condition (b) also nicely excludes making certain muscles move and sending certain nerve impulses, for even if the agent is a knowledgeable physiologist who knows exactly what nerve impulses and muscle flexings take place when he raises his right arm, his ability to produce these depends upon his cause-and-effeet knowledge. For in this case he knows that raising his arm would be the effect of these nerve Impulses, and muscle movements. The ability to execute certain basic acts depends on such processes, but we do not have to know about these processes to raise our arm. Instead, we merely raise our arm. 237 We are now in a position to understand condition (a) of Goldman's definition of a basic act-token, viz., property A is a basic act-type for S (at t). Condition (b), viz., S ’s exemplifying A (at t) is caused, in the charac teristic way, by an action plan of S,is a bit more difficult to explicate, however. The notion of an action-plan can be explained in terms of the notions of an agent's projected action-tree and his action wants. An action-tree is simply an act-diagram, and an act-diagram is a diagrammatic representation of the structural relationships between acts, i.e., of the genera tional relationships between acts. Thus, the following would represent a typical act-diagram: 238 Giving Frank Adams Giving his op a heart attack ponent a heart attack Checkmating Checkmating his Frank Adams opponent Moving his queen to king- knight-seven Frightening away Moving his queen a fly Moving his hand In this diagram. S's moving his hand causally generates both the act of frightening away a fly and also causally generates the act of moving his queen. The act of moving his queen augmentally generates the act of moving his queen to king- knight-seven which, in turn, conventionally generates both the act of checkmating Frank Adams and the act of checkmating his opponent. The act of checkmating Frank Adams causally generates both the act of giving Frank Adams a heart attack and the act of giving his opponent a heart attack. And, likewise, the act of checkmating his opponent causally generates both the act of giving Frank Adams a heart attack and the act of giving his opponent a heart attack. The acts of checkmating Frank Adams and checkmating his opponent are drawn on the same level since they differ from each other only in that they contain different individual concepts of the same object. The same applies to the acts of giving Frank Adams a heart attack and giving his opponent a heart attack. A projected act-tree is simply such a diagram when ever it represents the acts that someone (the agent himself or somebody else) believes would be performed if the agent were to perform a certain sort of basic act. Hence, an agent's action-plan is simply his projected action-tree plus his action wants. Or, briefly, the action-plan of an agent is the desire of the agent to perform some act plus a set of beliefs to the effect that this act will be generated by (or be on the same level as) a single basic act performed by the a g e n t . 186 Equipped with the notions of a basic act-token, an action-plan, and an intentional act, it is now possible for Goldman to account for two troublesome cases frequently ls$Ibid., p. 56. encountered in the philosophy of action. The first case concerns the possibility of there being certain "behaviors" which can be exemplified by agents as either acts or mere happenings. Consider, for instance, the phenomena of coughing and grimacing. Sometimes these behaviors are exemplified by one and the same agent, as happens when he coughs as a result of dense smoke or grimaces as a result of the taste of some foul-tasting food. But, at other times the same agent may cough or grimace on purpose, as for instance when the agent wishes to appear sick in order to escape some disagreeable task or when he wishes to insult his hostess. The difference between those cases in which these behaviors are exemplified as mere happenings and those cases in which they are exemplified as acts is simply that in the latter cases they are caused by some appropriate set of wants and beliefs, while in the former cases they are not. The definition of a basic act-token can also be used to distinguish between those cases where behaviors are exempli fied by an agent in a non-basic way and those cases in which the behavior is exemplified by the agent in a basic way. Thus, an agent’s simply raising his right hand is a basic act because it is not generated by any other act that the agent performs, while the agent’s raising his right arm by lifting it with his left hand is not a basic act-token because it is causally generated by some other basic act 241 that the agent has performed, i.e., raising his left hand. At this point I believe that it is possible to observe a glaring inconsistency between conditions (a) and (b) of Goldman's definition of a basic act-token. For condition (b) requires that a basic act be caused, in a certain characteristic way, by the agent’s prospective action plan. But, we have seen that, according to Goldman, an action- plan is a function of both the agent's wants and beliefs and, in particular, the agent's beliefs about what acts will be generated as a result of the agent’s performing some single basic act. Yet, condition (a) of the definition of a basic act-token requires that the property exemplified by the agent in performing a basic act (at t) be a basic act- type for the agent (at t), and condition (a) of the definition of a basic act-token requires only that agent's exemplifying that property (at t) result from the agent's wanting to exemplify that property (at t}. This inconsistency can also be seen in another way. It will again be recalled that Goldmari's definition of an in tentional action requires that it either be caused "directly" by the agent's prospective action-plan or its being level generated in a fashion conforming to the agent's prospective action-plan. Now the definition of an intentional action appears to make good sense if we are concerned with nonbasic 242 actions. But clearly all generated acts must have as their initial source some basic act, e.g., one cannot perform the act of signalling without performing some basic act or other. The question which naturally arises here then is whether the basic act itself is intentional or not. On Goldman's analy sis, the basic act itself must be intentional.187 But on Goldman’s account of an action-plan, an action-plan is a function of both the agent's wants and beliefs. But in spite of the fact that generally people perform simple bodily acts such as raising an arm in order to perform some other act, it sometimes happens that a person simply raises his arm without intending thereby to bring any other act to pass, i.e., he simply raises his arm because he wants to. Now it would appear that Goldman dismisses certain cases of salivating and fidgeting as intentional acts on the grounds that, in these cases, the relevant beliefs have no real causal role in the production of the salivating and fidgeting.188 And he likewise claims that except in the case of the per formance of a basic act both beliefs and desires are causes •*-8^In attempting to account for the difference between a (mere) bodily movement and a bodily action, Goldman opts for what he calls a traditional, though much castigated, answer. Namely, that for S's raising his hand to count as a basic act-token, it must be intentional, and for it to be inten tional, it must be caused by the agent's wants and beliefs. See p. 71. 188Ibid., p. 62. 243 of intentional acts. But clearly/ if, as Goldman claims, all basic acts are intentional then the definition of intentional action presented above cannot be considered to be the appropriate sense for intentional basic acts. For in spite of the fact that Goldman's definiton of an intentional action contains the possibility that An = 1, this possibility seems to be inconsistent with the remaining part of the defi nition. For in the case of a simple single basic act, there appears to be no necessity of the agent's having an action- plan at all. In order to save Goldman's definition of an intentional action and his thesis that all basic acts are intentional actions, we shall be forced to grant that when ever an agent performs the simple single basic act of raising his arm, there is some belief or set of beliefs which is causally relevant to the agent’s act of raising his arm. But the difficulty with this proposal is specifying exactly what it is that the agent must believe in this case, e.g., the agent wants to raise his right arm and believes that raising will level-generate his raising his right hand. "That hand," "That piece of flesh," etc., all seem to be mere fabricated cases, and the only thing that seems to have the required sense is "That is my right hand." 244 I believe that these difficulties can be successfully resolved by reformulating Goldman's definitions of inten tional action and basic act-token so as to allow for the possibility that a basic act may be caused either by a want alone or by wants and beliefs together, and the possibility of two kinds of intentional actions, intentional basic acts and intentional nonbasic acts. The resulting definition of a basic act-token would be as follows: S's exemplifying property A (at t) is a basic act-token if and only if (a) property A is a basic act-type for S (at t); (b) S's exemplifying A (at t) is caused, in the characteristic way, either by S's wanting to do A or by an action-plan of S; and (c) S's exemplifying A (at t) is not level generated by anything else satisfying clauses (a) and (b), except possibly by augmentation generation. Goldman's definition of an intentional act would become the corresponding definition of an intentional nonbasic act. The revised definition of an intentional basic act-token would be the same as the above, except that condition (b) would read: (b) S's exemplifying A (at t) is caused, in the characteristic way, by S's wanting to do A. However, this problem seems minor when compared to the obscurity which surrounds the phrase "in a certain character istic way" which appears in both the definition of intentional 245 action and condition (b) of the definition of a basic act- token. For unless something can be done to clarify what is meant by this phrase, the precise manner in which an agent's beliefs and wants produce his basic acts will re main a mystery. Goldman claims that he does not have a fully detailed analysis of what this phrase is supposed to entail, and sometimes he simply states that this is mainly a problem for the special sciences, perhaps one calling for extensive neurophysiological investigation, rather than a problem to be resolved by philosophical analysis. In another place, Goldman, following a suggestion of D.O. Hebb, explains that intentional or purposive behavior may be distinguished from nonvoluntary and nonpurposive behavior in terms of different paths from receptors to effectors. That is, the suggestion is that intentional and purposive behavior depends upon certain complex "closed loop" connections in the cortex, while reflexive and non voluntary behavior do not involve these connections. Hence, the phrase; "caused in a certain characteristic way" is 189lbid., p. 56. 246 supposed to refer to the kind of causation which involves these complex "closed loop" connections in the cortex. Reflex actions like grimacing at the foul taste of some food are supposed to be caused by a "straight-through" connection between receptors and effectors without involving these "closed loop" connections in the cortex. Thus, the foul taste of the food leads directly, via "straight- through" connections, to the contraction of the face muscles. Secondly , since a case of salivating is not a voluntary act yet may involve the agent's desire to eat, some way of explaining the salivation must be found. Goldman suggests that in this case the desire (the cortical event) does not cause the salivation direcbly, but only indirectly by causing the activation of the autonomic nervous system which, in turn, causes the salivation. Finally, the case in which an agent's desire to attract the speaker's attention which causes him to be nervous, tense, and fearful, and hence to perform the involuntary act of fidgeting in his chair is perhaps to be accounted for in the following way. The nervousness, tension, or fear causes an excitement of the sympathetic nervous sys tem, which, in turn, causes an increase in the secretion of adrenalin into the bloodstream. This increase of adrenalin has the effect of lowering the electrical resistances of the 247 synapses causing them to fire with much smaller amounts of electrical charge than normally required. Because of this increase in the rapidity of neuron firings, muscle move ments also occur with a greater frequency, and this is the fidgeting. All of this is, of course, rather conjectural as Goldman admits. But what is surprising is not their con jectural character, but rather, that Goldman would have made them in the first place seeing that he has explicitly stated that wants and beliefs are mental events or dispositions. 5.4. Hypothetical Accounts of Human Ability The concept of having the ability to perform an action certainly appears to be a dispositional concept, and one which, I believe, can be successfully analyzed in terms of Goldman’s notions of act-generation and basic act-type. It will be recalled that the first part of Goldman's definition of a basic act-type is this: "If S were in standard conditions with respect to A at t, then if S wanted to exemplify A at t, S's exemplifying A at t would result from this want." Nov/, the subjunctive, "If S wanted to exemplify a at t, S's exemplifying A at t would result from this want," is Goldman's interpretation of the sentence, "S can (cannot) exemplify property A at will," and is a causal subjunctive which refers to a dispositional property of the agent S. Hence, on Goldman’s analysis the causal conditional statement "If S wanted to do A at t, then S would do A at t," is a sufficient condition for and entails the truth of the statement, ,:S is able to do A at t. But such hypothetical or conditional analyses of human ability have been questioned by several contemporary phil osophers. For example, in his article, "Ifs and Cans," J.L. Austin argues against both G.E. Moore's claim that "I could have . . is to be translated as "I could have . . if I had chosen," and Nowel-Smith's claim that "I can . . is to be translated as "I will . . . if . . ." and "I could have . . ." is to be translated as "I would have . . . if . . Avoiding for the moment the complications involved in Austin's distinctions between the three senses of "cans" (i.e., the opportunity sense, the ability sense, and the "all-in" sense of opportunity plus ability), basically Austin accuses both Moore and Nowell-Smith of failing to distinguish between two different and distinct claims, the claim that statements containing "can" and "could have" are incomplete in meaning and require an "if"-clause to complete their meanings, and the claim that such statements cannot be 249 adequately analysed without an "if"-clause.1®® Hence, Austin claims, One view is that wherever we have can or could have as our main verb, an if-clause must always be understood or supplied, if it is not actually present, in order to complete the sense of the sentence. The other view is that the meaning of "can" or "could have" can be more clearly reproduced by some other verb (notably "shall" or "should have") with an if-clause appended to it. The first view is that an i£ is re quired to complete a can sentence; the second view is that an if is required in the analysis of a can-sentence. The suggestion of Moore that "could have" means "could have if X had chosen" is a suggestion of the first kind; but the suggestion also made by Moore that it means "should have if I had chosen" is a suggestion of the second kind.191 Austin believes that both of these theses are false, although it is difficult to see how they can be incompatible, as he claims they are, and at the same time be both false, unless they have the incompatibility of contraries. In any case, Austin argues that the thesis that statements containing "could have" and "can" are incomplete in meaning and need an "if"-clause to complete their meaning is false because the notion that there could be a language which contained verbs which could not occur without an "if"-clause attached to them would be senseless. Hence, he claims, For let the verb in question be t£ X: Then we shall never say simply "I X," but always "I X if I Y": but then also, according to the accepted rules, if it is 190J.L. Austin, "tfs and Cans," in The Nature of Human Action, ed. by Brand, op. cit., pp. 167-171. 191Ibid., p. 167. 250 true that "I X if I Y ," and also true (which it must surely sometimes be) that "I do, in fact, Y"; it must surely follow that "I X," simpliciter, without any i£ about it any longer.192 Unfortunately, it is difficult to see how this argument shows the first claim is false. For one thing, it is difficult to see why even if some verbs must be capable of appearing alone without an "if"-clause in a language, it follows that "can" and "could have" belong in this class. Austin's major objection to the second thesis, viz., that ability statements are analyzable in terms of state ments like "I will . . . if . . . ,11 or "I would have . . . if I had chosen (tried)," is that my being able to do A is consistent with choosing, trying, and wanting to do A, but failing to do A. Thus, he argues, Consider the case where I miss a very short putt and kick myself because I could have holed it. It is not that I should have holed it if I had tried: I did try, and missed. It is not that I should have holed it if conditions had been different: that might of course be so, but I am talking about conditions as they precisely were, and asserting that I could have holed it. There's the rub. Nor does "I can hole it this time" mean that I shall hole it this time if I try or if anything else: for I may try and miss, and yet not be convinced that I couldn't have done it; indeed, further ex periments may confirm my belief that I could have done it that time although I didn't.193 1®2Ibid., p. 168. 193Ibid., p. 169n. 251 Along the same lines, Austin criticizes Nowell-Smith’s argument that "I can . . ." means "I will . . . if . . ." The heart of Nowell-Smith1s argument is that there would be some thing ''logically odd" about asserting that X had the ability to do A, I had an opportunity to do A, and I had a motive for doing A, but nevertheless I did not do A. And Austin cor rectly shows that although this shows that if I have the abil ity, then, if I had the opportunity and the motive to do A, I will do A, it does not entail that if, when I have the op portunity and the motive to do A, I do A, then necessarily I have the ability to do A. Now, regardless of how valid these arguments are against Moore's and Nowell-Smith's proposals, they do not seem to touch Goldman's proposal. For in the first case, Goldman's proposal is only claimed to be a sufficient condition for saying that an individual can (is able to) do A at t. It does not claim that the statement that an individual can (is able to) do A at t is a sufficient condition for the truth of the subjunctive proposition, "if S wanted to exemplify A at t, S's exemplifying A at t would result from this want." In other words, Goldman is not claiming that "S can (is able to) do A at t" means "If S wanted to exemplify A at t, S's exem plifying A at t would result from this want." And in the 194ibid., pp. 174-175. 252 second place, Goldman's proposal is meant to apply only to basic act-types or basic act-tokens of an agent. Austin's counterexample, on the other hand, is an instance of a non basic act-type (token) since, so far as I know, holing a golf putt is not a basic action for anyone. Holing a golf putt is something which clearly depends on the agent's knowledge of causes and effects since the golfer must know what to expect when the head of the putter he is swinging makes impact with the ball. The more important issue here seems to revolve around Goldman’s claim that this subjunctive conditional is most correctly interpreted as a causal subjunctive which refers to a dispositional property of the agent. Again, Austin has argued in his "Ifs and Cans" that the "if" in "he can if he chooses" and "he could have if he had chosen" are non- causal, and that there are many other noncausal uses of "if" in connection with other verbs. Hence, he says, Generally, philosophers, as also grammarians, have a favorite, if somewhat blurred and diffuse, idea of an if-clause as a "conditional" clause: Putting our example schematically as "If p, then q," then it will be said that q follows from p, typically either in the sense that p entails q or in the sense that p is a cause of q, though other important variations are possible . . . But now, it is characteristic of this general sort of i£, that from "if p then q" we can draw the inference "If not q, then not p," whereas we can not infer either "Whether or not p, then q," or “q" simpliciter. For example, from "If I run, I pant" we can infer "if I do not pant, I do not run" (or, as we should rather say, "If I am not panting, I am not running"), whereas we can not infer either 253 "I pant, whether I run or not1' or "I pant" (at least in the sense of "I am panting") . . . These possibilities and impossibilities of inference are typical of the' if of causal condition: but they are precisely reversed in the case of "I can if I choose" or "I could have if I had chosen." For from these we should not draw the curious inferences that "If I cannot, I do not choose to" or that "If I could not have, I had not chosen to" (or "did not choose to"), whatever these sentences may be supposed to mean. But on the contrary, from "I can if I choose" we certainly should infer that "I can, whether I choose to or not" and indeed that "I can" period: and from "I could have, whether I chose to or not" and that anyway "I could have" period. So that, whatever this if means, it is evidently not the if_ of causal condition. This becomes even clearer when we observe that it is quite common elsewhere to find an ordinary causal conditional if in connection with a can, and that then there is no doubt about it, as for example in the sentence "I can squeeze through if I am thin enough," which does imply that "If I cannot squeeze through I am not thin enough," and of course does not imply that "I can squeeze through." "I can if I choose" is pre cisely different from this. Nor does can have to be a very special and peculiar verb for ifs which are not causal con ditional to be found in connection with it: all kinds of ifs are found with all kinds of verbs. Consider for example the if in "There are biscuits on the sideboard if you want them," where the verb is the highly ordinary are, but the if[ is more like that in "I can if I choose" than that in "I panted if I ran": for we can certainly infer from it that "There are biscuits on the sideboard whether you want them or not" and that anyway "There are biscuits on the sideboard," whereas it would be folly to infer that "If there are no biscuits on the side board you do not want them," or to understand the meaning to be that you have only to want biscuits to cause them to be on the sideboard.I” 195ibi.d. , pp. 164-165. 254 In response to this criticism of Austin's, Goldman replies that, although the "if" in "he can if he chooses" or in "he can if he wants" may be non-causal, it does not follow that the "if" in "he will if he chooses," or in "he would if he 196 chose," or in "he would if he wanted" are non-causal. Although Austin does not explicitly treat attempted analyses of ability in terms of other notions besides choosing (e.g., wanting, desiring, wishing, etc.), Keith Lehrer in his article "Ifs, Cans, and Causes,"197 has produced an argument which he claims is equally telling against such alternative analyses. But the difficulty with that argument is that, again, it does not apply to Goldman's proposal since he is not claiming that l!S can do A at t" is equivalent to "If S wanted to exemplify A at t, S's exemplifying A at t would result from this want." Goldman is also correct in his claim that Austin never shows that the "if" in "he will if he chooses," and in "he would if he chose" are non-causal. For one thing, Austin never explicitly treats these expressions. The closest he Alvin Goldman, A Theory of Human Action, op. cit., p. 64. l^'Keith Lehrer, "Ifs, Cans, and Causes," in The Nature of Human Action, ed, by Myles Brand (Glenview, Illinois: Scott, Foresman and Company, 1970), pp. 179-181. 255 comes to it is his treatment of the expression “I shall if I choose." And with regard to this expression he argues, And now for something about "I shall if I choose" - what sort of if have we here? The point to notice is, that "I shall" is not an assertion of fact but an expression of intention, verging towards the giving of some variety of undertaking: and the if, consequently, is the if not of condition but of stipulation. In sentences like: I shall / marry him if I choose. I intend / to marry him if I choose. I promise / to marry him if he will have me. the if-clause is a part of the object phrase governed by the initial verb ("shall," "intend," "promise"), if this is an allowable way of putting it: or again, the if qualifies the Content of the undertaking given, or of the intention announced, it does not qualify the giving of the undertaking. 198 Hence, Austin seems to be claiming that the if in "I shall if I choose" is like the if in the sentences "I intend / to if I choose," and "I promise to if" in that it is not the if of causal condition (or any condition). However, Keith Lehrer has criticized Austin on this point. Lehrer argues that Austin made a mistake in comparing "I shall if I choose" with "I intend to if I choose." Ac cording to Lehrer, expressions of the form "I shall . . . if" are not typically used to express intentions, although some times they do express an intention to do something with a I98Keith Lehrer, "An Empirical Disproof of Determinism?" in Determinism,' Free Will, and Moral Responsibility, ed. by Gerald Dworkin (Englewood Cliffs: Prentice-Hall, The., 1970), pp. 186-187. 256 certain stipulation (e.g., "I shall run for office if my brother gets elected). In fact, the expression ”1 shall if I choose" is usually used' where there is some doubt whether the speaker has any choice, and the use of the expression has the effect of cancelling this suggestion and affirming that he has the choice. Hence, if a young lady says, "I shall marry him if I choose,” she is expressing the intention to do so on condition that she chooses. Lehrer has a valid criticism here. Austin has not suc cessfully shown that the if_ in "I shall if I choose" does not express a condition, and hence, not a causal condition. 199 It is interesting to note that in the same place that Professor Lehrer argues against Austin's claim that the ij£ in "I shall if I choose" is not the if of causal condition or any condition whatsoever, he presents another argument to show that the statement that a person can do something is not entailed by any causal conditional which asserts that he v/ill do the thing in question if certain conditions are 199jt is conceivable that someone may want to attempt to defend Austin on this point by arguing that Lehrer's objec tion is based upon another instance of the peculiarity of first person utterances, or by arguing that there is some significant difference here between the verbs "shall" and "will." But I must confess that I do not knov; how these arguments would be stated. 257 satisfied.200 Basically/ his argument is that we can produce two statements which are compatible with the causal condi tional "S will do A if condition C obtains" but rather than entailing that S can do A, they entail that S cannot do A. These two statements are: "S cannot do A if condition C does not obtain," and "Condition C does not obtain." For the sake of facilitating discussion of this argument, let us label the relevant propositions in it in the following manner: (P) S will do A if condition C obtains, (Q) S cannot do A if condition C does not obtain, (R) Condition C does not obtain, (T> S cannot do A. Now, according to Lehrer's argument, Q and R are compatible with P. Q is compatible because it is logically possible that condition C which is asserted by P to be a causally sufficient condition for S's doing A may also be a necessary condition for S's being able to do A. And R is compatible with P because it is also logically possible that condition C may fail to occur. But Q and R entail T. And if P, Q, and R are jointly consistent and entail T, then neither 200Keith Lehrer, "An Empirical Disproof of Determinism?" op. cit., p. 190. 258 P, Q, nor R alone can entail the denial of T; viz., S can do A. Therefore, "S can do A ” is not entailed by any causal conditional of the form "S will do A if C." Goldman is not unaware of this argument and attempts to refute it. He suggests that the argument may be refuted by a careful discussion of temporal references and/or counterfactual conditionals, but he does not pursue this line of argument. Instead, he attempts a reductio ad absurdum argument against Lehrer. His argument is as follows: It is evident that there are some statements that can correctly be analyzed in terms of conditionals. A paradigm case of such a statement is P': "X is soluble in water," which is analyzable as Q': "X will dissolve if it is immersed in water," or, more accurately, as Q '': "X would dissolve if it were immersed in water." But if Lehrer's argument is sound, we can use an argument of the same form to "prove" that P' can not be correctly analyzed as Q 1 (or as Q 1'). For consider statements R ' : "If X is not immersed in water it is not soluble." and S': "X is not immersed in water." 259 R* and S' are jointly compatible with Q' (and Q11). For let X be a piece of sugar, and imagine a magician such that if X is not im mersed in water he changes its molecular structure so as to make it non-soluble, where as if X is immersed in water it dissolves. Under these circumstances, it seems that R* and S' are jointly compatible with A' (or Q 1’). But R 1 and S’ entail the negation of P 1. Hence, if all of this is correct, then Q 1 (or Q'1) does not entail P 1. We see, therefore, that Lehrer’s form of argument leads to absurd conclusions, e.cr. , that P 1-cannot be analyzed as Q' (or as Q'').20-1- Goldman's implication that Lehrer’s argument can be shown to be unsound by adding the necessary time references is not prima facie justified. Apparently the procedure here would be to show, by the use of such precise time references, that although it may be true to say that a person S is able to do A at one time, it may be false to assert that he can do A at some other time. But it is not clear how inserting such time references into Lehrer's argument will show how it is unsound. For, supposing that the proposition "S can do A at time t" is the proposition which is to be analyzed, we could rewrite Lehrer's argument 20^Alvin Goldman, o p . cit., pp. 199-200n. I have taken the liberty here of adding 's and ’’s to Professor Goldman's P*s, Q's, R ’s and S ’s so that they will not be confused with the P ’s, Q ’s, R ’s, and S ’s in Lehrer’s argument. 260 as follows: (P ' ' ') S will do A at time t if condition C obtains at time t, (Q1'1) S cannot do A at time t if condition C does not obtain at time t, (R''') Condition C does not obtain at time t, (T11') S cannot do A at time t. Now, it seems to me that by reformulating the argument in this way, we have not succeeded in destroying it at all. My final criticism of Goldman’s refutation by analogy concerns the possible justification for supposing that X is still sugar after the magician has changed its molecular structure in the fashion that Goldman suggests. For so far as I know, there is no definition of solubility in terms of a thing’s molecular structure. Hence, if this is correct, Goldman's R 1 and S' are not jointly compatible with Q' (and Q ” ). However, this last criticism seems to be capable of working both ways. That is, it may be claimed that if the criticism is applicable to Goldman's argument, it is also applicable to Lehrer's original argument, in that it is also possible to claim that the person S mentioned in proposition P is not the same person S which is mentioned in propositions Q, R, and T. And, if this is the case, Lehrer's argument is invalid since it shifts the meaning of that term. I know of 261 no satisfactory way of meeting such a rebuttal. One may argue that, whereas there are only a very few varieties of criteria for being a lump of sugar, there are, on the other hand, a great variety of criteria for personal identity. But it would be difficult to see how this in itself would destroy the criticism. For even if there are a larger number of criteria for being a person than there are for being a lump of sugar, there is no reason to suppose that these criteria are indefinitely large. It is perhaps extremely difficult to specify the exact point at which someone ceases to be the same person and becomes another person or something else, but it does not follow that there are absolutely no ways of telling when such a change has taken place. Of course, Goldman is aware of the fact that so far he has only offered a sufficient condition for S's being able to perform a basic act at t, and he attempts to amend this by providing a definition of ability which is intended to account for all acts. Hence, he provides the following definition: S is able, at t, to do A at t if and only if: There is a basic act A^ for S at t and a set of conditions C* such that (a) S is in standard conditions with respect to A-^ at t, and (b) either A=A. or S's doing A^ at t would generate S's doing A at t, 202 202Ibid. , p. 201. 262 This definition of ability contains two notions which are also central to Goldman's definition of a basic act-type and a basic act-token, i.e., the notion of standard con ditions and the notion of level generation. However, both of these notions have already been explained and should not engender any great misunderstandings here. CHAPTER VI WANT AND BELIEF - EXPLANATIONS AND MECHANISM In the preceding chapter I argued that G. E. M. Ans- combe's attempt to define intentional actions suffered from some serious defects, one of these being the sup position that the agent knows the actions he performs, non-inductively and non-observationally. Next, it was argued that Anscombe's attempts to justify her claims about how a person knows his own intentions in acting was based upon a weak analogy and the mistaken assumption that some avowals of intention cannot be questioned. In regard to the latter of these it was argued that although there are some criteria which we commonly employ in determining the presence or absence of intentions, none of them were either necessary or sufficient. In view of the failure of these attempts to define intentional acts, a preliminary justification of a causal account of intentional action in terms of wants and beliefs was offered. Next, Goldman's account of intentional action was presented, and I argued that, barring an inconsistency in his definition of a basic act-token which could easily be avoided by redefining a basic act-token, Goldman's 263 264 account was completely satisfactory. Finally, I defended Goldman's hypothetical/dispositional account of human ability against the criticisms of J. L. Austin and others. In this chapter I will explore some of the unresolved difficulties with Goldman's theory of human action in cluding the claim that the explanations of human actions offered by the behavioral sciences are not in competition with our everyday explanations of human actions in terms of wants and beliefs. 6.1. Wants and Beliefs Although I have been arguing that our common-sensical notion of a human action can be accurately analyzed in terms of Goldman's definitions of act-generation, and a slightly revised definition of basic act-token, and in tentional action, his failure to provide a similar in- depth analysis of wants and beliefs seriously detracts from the overall plausibility of his theory. For if all acts are caused by wants and beliefs, then in order to fully explicate the notion of a human action some account of the status of wants and beliefs is in order. One would like, after all, to know precisely what wants and beliefs are, and Goldman has failed to provide any com prehensive analysis of wants and beliefs. One distinction which seems uncontroversial is the distinction between occurrent or aroused wants and standing 265 or latent wants.203 consider a man who is in the process of cleaning the windows of his house. Suppose that he is doing so with the thought of finishing it before five o'clock, and consequently, he is cleaning the windows as quickly as possible. The thought of finishing before five o'clock occurs to him during this period. It is a datable event or process. Thus, it seems appropriate to say of him that during this period that he has an occurrent or aroused want to finish the windows by five o'clock. Now, suppose that during the same period it is also true that he wants to be well liked by people. But his wanting to be well liked by people is not the same kind of want as his want to finish cleaning the windows by five o'clock. For no such thought occurred to him during this period. Hence, in attributing such a want to him we are not saying that at this particular time he is entertaining any thoughts on the subject of other people's attitudes toward him, but rather that he has a latent or standing want to be well liked by other people. And in attributing such a latent or standing want to him we are saying that he has a disposition to have oc current or aroused wants to this effect. It would appear that a similar distinction can be drawn between standing and occurrent beliefs also. 203C . William P. Alston, "Motives and Motivation," in The Encyclopedia of Philosophy, V, 399-409, and Alvin Goldman, A Theory of Human Action, op. cit., p. 86. 266 There would appear to be nothing in the logical analysis of want and belief sentences which would rule out construing at least some of them as dispositions. The central difficulty in attempting to provide a logical analysis of want and belief sentences is the teleological character of wants and beliefs. That is, wants and beliefs, as they are ordinarily understood, are thought to possess the characteristic of being "for" or ’'about" something, e.g., John wants to study physics or John believes that it is snowing. But the difficulty with such "objects of desire or belief" is that frequently they are nonexistent, false, or self-contradictory. That is, one can want something even when the thing wanted is nonexistent, or one can want to do something without ever doing it, or one can believe something which is false or even self-contradictory. These difficulties are, of course, serious, but they are not of the sort which are insurmountable. Israel Scheffler has presented an inscriptional analysis of belief and want statements which, I believe, takes into account all of these difficulties. ^04 204see Israel Scheffler, The Anatomy of Inquiry (New York: Alfred A. Knopf, 1963), pp. 88-111. 267 Scheffler's inscriptional analysis of belief and want sentences resulted, in part, from his attempts to show that a deductive explanation of performances could be given in terras of the agent's desires and beliefs and a generali zation connecting the desire and belief with the performance. The precise difficulties involved in such a project can be readily seen from an examination of the following rather artificial and simplified attempt at providing a deductive explanation of an agent's performance in terms of his beliefs and desires; (1) John desires John's qualification for entrance to medical school. (2) John believes that John's qualification for entrance to medical school is con tingent on John's choice of a premedical course. (3) Whenever someone desires something, be lieving that it is contingent on something else, he performs this latter thing. (4) John performs John's choice of a pre medical course. Now, the generalization connecting the desire and belief with the performance (premise number (3) in the above argument) cannot be construed in such a way that the desire and belief are represented as two-place predicates. For in symbols, (3) would become: 268 (x) (y) (z) ({Desires xy • xB (Counting yz))0 Pxz) That is: For all x # y, and z, if x desires y and believes that y is contingent on z, then x perforins z. The difficulty comes with determining the range of 'y1 in the desire con text and the range of 'y' and 'z' in the belief context. For given that the "object of John's desire" may not exist (e.g., "John's qualifying for entrance to medical school" may not occur), premise number (1) will be false. That is, (9y) (John desires y) is false, and this would have the unsavory consequence of destroying the value of the argu ment. Hence, a desire cannot be construed as a two-place predicate. The same difficulty arises with the two-place rendering of the belief since as 'y' and 'z' appear within the belief context, y may be unrealized and z not performed. Scheffler's solution of these difficulties consists of replacing the two-place predicates with operators that express special relationships between agents and inscrip tions. Inscriptions are physical objects which can be counted on to exist despite the fact that desires are frequently thwarted, beliefs are frequently false, per formances are frequently not accomplished, and agents are frequently irrational. The new operators are "desiring true," represented by DTr; "believes true," represented by BTr; and "makes true," represented by MTr. The above 269 deductive argument can now be reformulated as follows: (20) (c/x) (That (John-qualifies-for-entrance-to- medical-school)x • DTr Jx) . (21) (-Tx) (That (If-John-qualifies-for-entrance- to-medical-school-then-John-chooses-a-pre- medical-course)x • BTr Jx). (23) (x) (v) (z) (w) (Kzvw • DTr xv • BTr xzoMTr xw) . (24) John chooses a pre-medical course.205 "Kzvw" in (23) means that z is a conditional formed out of v and w , in that order. In words (23) says: For all x, v, z, and w, if z is a conditional formed of v and w, and x de- sires-true v, and x believes-true z, then x makes-true w. And since the variables in the above argument range only over inscriptions and inscriptions are existent physical objects which exist in spite of their falsity or self contradictor iness, an agent may "desire true" some inscrip tion without producing it, possessing it, wishing to possess it, being aware of it, or even understanding the inscription. Likewise, an agent may believe true some inscription even if it is the case that the inscription itself is false or self contradictory. But, of course, it would defeat the whole purpose of the analysis to allow that an agent can make true 205Ibid., pp. 103-106. The numbers are Scheffler’s. 270 an inscription so MTr ____ is true if and only if the inscription is in fact true. It should be mentioned, however, that there is at least one serious logical difficulty with the inscriptional analysis of beliefs and desires. For it will be recalled that the inscriptionalist analyzes a belief sentence like: (1) "John believes that Ortcutt is a spy" as (2) "John believes-true 'Ortcutt is a spy.1" Nov/, the difficulty is that it would seem that by virtue of existential generalization we may validly infer: (3) " Ox) (John believes that x is a spy) " from (1). But if we attempt to do the same on the proposed analysans of (1), viz., (2), the result is: (4) ” (H|x) (John believes-true 'x is a spy')." And (4) is supposedly a falsehood preceded by an irrelevant quantifier.20^ Hence, the inscriptional analysis of (1) does not appear to have the same intuitively felt conse quences as (1), and cannot therefore be considered to be the correct analysis of belief sentences. 20®Willard van Orman Quine, From a Logical Point of View, 2nd ed.; (New York: Harper and Row, Publishers, Inc., 1961), pp. 145ff. 271 This objection, however, fails to note, as Quine has so skillfully pointed out,2C^ that existential generaliza tion may or may not be performed upon (1) itself, depending upon how it is construed. That is, 'believes that* in (1) may be construed either as referentially opaque or referen- tially transparent, i.e., the occurrence of 'Ortcutt* in (1) may be interpreted as either occupying a purely refer ential position or not. A term occupies a purely referential position if and only if it is subject to the substitutivity of identity. If ’believes that1 in (1) is construed as referentially opaque, i.e., the occurrence of 'Ortcutt* is interpreted as not occupying a purely referential position, then we cannot validly perform existential generalization on (1) in the manner of (3). For to suppose that we may validly perform existential generalization on (1), con strued in this fashion, given that Ortcutt=Smith and that John does not believe that Smith is a spy, is to suppose that "John believes that Smith is a spy" is in fact true, whereby, by hypothesis, it is clearly false. If, on the other hand, 'believes that* in (1) is construed as refer entially transparent, i.e., the occurrence of 'Ortcutt' is interpreted as occupying a purely referential position, then we may validly perform existential generalization on (1). 2°7 Dillard van Orman Quine, Word and Object, (Cambridge, Massachusetts: The M.I.T. Press, 1960), pp. 142-146. 272 Clearly, the reason why existential generalization cannot be validly performed on the inscriptional analysis of (1), viz., "John believes-true 'Ortcutt is a spy1" is simply that in it 'Ortcutt' is construed as not occupying a purely referential position. Quotation typically gives rise to non-referential positions. There is, however, a way of construing the proposed analysans so that 'Ortcutt1 does occupy a referentially transparent position. In this case, instead of (2) above, the inscriptional analysis would be: (5) "John believes-true 1x is a spy' of Ortcutt."208 Now, since ’Ortcutt" does occupy a referentially transparent position in (5), i.e., the believes-true context is no longer referentially opaque, we may validly quantify into it. Thus, (6) "$y) (John believes-true 'x is a spy' of y)" which saves the inscriptional analysis. In addition to this, it is questionable whether the objection to the inscriptional analysis that we have been considering is itself capable of withstanding critical evaluation. For, clearly the objection cannot get off the ground unless the inference from (1) to (3) is justified. It is easy to see, however, that this is not always the 208 Ibid., pp. 168, 212. 273 case. Consider, for instance, (7) "John believes that Santa Claus lives at the North Pole." The existential generalization of this,' viz., (8) 11 (3x) (John believes that x lives at the North Pole)" is clearly false. Of course, (9) "John believes that Ox) (x lives at the North Pole)" is true, but here both '3x' and ’x' occur within the opaque believes context.209 Hence, there would appear to be nothing in the logical analysis of want and belief sentences which would tell against construing wants and beliefs as either dispositions or occurrent events. But, it will be recalled that Melden has leveled several criticisms against dispositional accounts of behavior. These criticisms are: (1) people can and do sometimes act out of character, (2) attempts to avoid this criticism by appealing to "standard or normal conditions" are unsatisfactory because standard or normal conditions cannot be defined, (3) attempts to avoid this criticism by appealing to "interfering circumstances" makes the dis positional account useless and trivial and (4) dispositional accounts are unhelpful because of the obscurity of the notion of a contrary-to-fact conditional. 209Ibid., p. 167. 274 But these criticisms are not sufficient to show that a large part of a person's behavior cannot be accounted for in terms of dispositional wants and beliefs plus some "triggering " events. For in the first place, what is meant by a person's character is extremely vague. Our common-sensical attributions of character traits to in dividuals are more times than not based upon inadequate evidence and the character traits that we common attribute to individuals can hardly be said to be well-defined. Can we, for instance, even pretend that when v/e say that a man is polite, or kind, or irritable, and so on, that we have succeeded in precisely characterizing him? It is indeed hard to believe that all attributions of dispositional wants and beliefs must be so imprecise as these are. And in the second place, Melden's criticisms simply ignore the pos sibility of time-dependent dispositions. To attribute a dispositional property to a thing is not to say that that thing will necessarily exemplify that property for all time, nor is it to say that that thing will exemplify that property for a long time to come. Thus, to say that something is magnetic, highly elastic, brittle, or irritable is not to say that it will exemplify those properties always. And finally, the criticism based upon the general obscurity of the notion of a contrary-to-fact conditional cannot be taken seriously since it is a difficulty that applies equally not 275 only to dispositional properties of human agents, but to every other dispositional property as well. And the mere fact that no one to my knowledge has come up with a com pletely adequate analysis of contrary-to-fact conditionals is not sufficient reason for supposing that no such analysis is possible. perhaps Melden's reluctance to accept the "triggering" of dispositions along with events as possible causes of human actions springs from the supposition that event causation is incompatible with agent causation. That at least one contemporary action theorist has believed that there is some incompatibility here is evident from what Richard Taylor has to say on the subject. Hence, he writes, It is plain that, whatever I am, I am never identical with any such event, process, or state as is usually proposed as the "real cause" of my act, such as some intention or state of willing. Hence, if it is really and unmetaphorically true, as I believe it to be, that I sometimes cause something to hap pen, this would seem to entail that it is false that any event, process, or state not identical with myself should be the real cause of it. This view, I believe, is engendered by the perception that ordinarily the locus of human actions is a human agent. Thus, in his book Human Acts, Eric D'Arcy says, As a general rule, an action is called an act only when it can be described in a proposition with a 210 Richard Taylor, Action and Purpose, op. cit., p. 111. 276 personal subject; the actions of signing a cheque or killing a rival are acts, for one can say, "I signed the cheque," or "He killed his rival"; but the beating of the heart and the working of the liver are not acts: one cannot say, "I beat my heart," or "I worked my liver."211 And a little later he suggests another version of this general rule: A slightly different version of the test is to say that one may substitute "act,: for "action" only when the action may be spoken of as "my action*': for example, the action of signing a cheque, or killing a man, may be called "my action," and is an act; the beating of the heart cannot be called "my action," and is not an act.212 These rules clearly suggest that the locus of a human action is always a person considered as a whole. Ordinarily, human actions are not predicated of the individual parts of a person. Thus, although the beating of a human heart or the workings of a human liver are classifiable as actions, they are not ordinarily classifiable either as human acts or actions. But clearly there is no incompatibility between event- causation and object-causation, and since agents are kinds of objects, there is no incompatibility between event- causation and agent causation. To say that an object caused 21lEric D'Arcy, Human Acts: An Essay in Their Moral Evaluation (Oxford, England: Oxford University Press, 1963), pp. 6-7.’ 212Ibid., p. 7. 277 something to happen is just to say that there is some state of that object or event involving that object that causes, or was a partial cause of, the event in question. That this is so is clear in the case of physical objects. Thus, it is not the dynamite that caused the building to collapse but the explosion of the dynamite that caused the building to collapse. It is not the bell ringing that caused John to jump, but the bell being heard by John that caused him to jump. Likewise, it is clear that the event of John’s flipping the light switch is the cause of the light going on, and this is in no way incompatible with the assertion that John was the cause of the light's going on. And, finally, if it is true, as I have been arguing, that all basic acts are caused by the wants and beliefs of the agent, then not only is there no incompatibility between agent- causation of acts and event-causation of acts, but also agent-causation can be explained in terms of want-and-belief causation. Likewise, there is no incompatibility between ex plaining a phenomenon in terms of dispositional properties of an object and an explanation in terms of some structural micro-properties of that object. Thus, the ability of one object to attract a body of opposite charge and to repel one of like charge is explainable by a surplus (negative 278 charge) or a deficit (positive charge) of electrons within the body. It may be, however, that such dispositional talk is, as Pap and Feigl suggest, •LJ'merely characteristic of the pretheoretical stage of a science. For consider what is involved in attributing the property of being magnetic to a bar. Now, it is by no means inconceivable that the bar has this particular property in virtue of some of its structural microproperties. But, so far as I know, no one is yet in a position to explain exactly what it is about the internal structure of this bar which is responsible for its exhibiting the property of attracting small ferrous metal objects. Likewise, it was once common to say that glass breaks because it is brittle, or water freezes be cause it has a disposition to solidify at 0°C, sugar dissolves because it is soluble, and so on, while now we have enough knowledge about the internal molecular structures of these things to offer other, more satisfactory explan ations of these phenomena. In this sense, dispositional concepts may be viewed as simply promisory notes to be filled in as our knowledge of the internal microproperties, of objects increases. Since standing wants and beliefs 213Cf. Arthur Pap, An Introduction to the Philosophy of Science (New York: The Free Press of Glencoe, 1962), Chapter IS-. 279 are attributed to agents as a whole, it is quite possible that,as our knowledge of the neurophysiology of the body increases, we will no longer have any need of the concept of standing wants and beliefs to explain human behavior. 6.2. The Supposed Mental Status of Wants and Beliefs The exact ontological status of certain items of awareness has constituted one of the most hotly-debated issues in the philosophy of mind for several centuries. Goldman, however, definitely sides with the mentalists on this issue, for he clearly believes that at least occurrent wants are mental events or episodes, and although he never explicitly says so, it is clear that he conceives of occur rent beliefs as mental events or episodes also. He seems to have been led to this view from the consideration that an occurrent want is "a 'going on' o r 'happening' in conscious ness." But, as I have argued in reference to the claim that the essential feature of a human agent and hence of the notion of acting from a reason is the characteristic of consciousness, that this supposed characteristic of con sciousness may prove in the end to be little more than a mere prejudice. Goldman's arguments for the supposed mental status of occurrent wants are, unfortunately, extremely weak. His first argument simply consists in pointing out that it is a mistake to argue, as he interprets Gilbert Ryle as 280 arguing, that since wants are not feelings or sensations, they are not (mental) events or episodes. This is, of course, true, but showing this does ,not show that occurrent wants are mental events or episodes. As far as I can determine, Goldman has only one posi tive argument to demonstrate the mental status of wants and beliefs. Thus, he argues, It is true, I think that we do not normally need evidence to know our occurrent wants or beliefs, that we do not have to infer them from other sources. Nor do we have to focus or concentrate on our wants to have a sort of knowledge of them, because a kind of implicit knowledge is already contained in them, as it were.2-1-4 This argument is not, however, relevantly different from the argument (already considered and rejected in Chapter III) that reasons cannot be causes because they are known in a relevantly different way than the way in which causes are known. In any case, even if Goldman's arguments do not es tablish that occurrent wants and beliefs are mental events 214 Alvin Goldman, A Theory of Human Action, op. cit., pp. 95-96. 281 or episodes, it certainly does not follow that we are under any obligation to say that they are. I have already indi cated that as far as I know, there is no way of demonstrating that there either is or is not any special class of things called mental things. However, it would appear that the greatest burden of proof falls on the shoulders of those who claim that there are mental things, since most of them at least agree that there are physical things. It is primarily for this reason that I will not be concerned with philosophical behaviorism in the following discussion of behaviorism as a rival explanation of human actions. Another reason for my lack of concern with philo sophical behaviorism springs from the fact that analytical behaviorism or the view that all psychological sentences, i.e., all sentences employing psychological or mentalistic terms, can be analyzed into sentences using only terms that refer to some kind of bodily behavior seems to be subject to the difficulty that no such behavioral analysis has as yet been successful. as j will shortly argue, the view that at least wants and beliefs may be successfully identified with certain neural events, states, and/or structures seems to be a far more.plausible position to take. 215Cf. r .m . Chisholm's claim that belief-sentences cannot be so analyzed. R.M. Chilholm, Perceiving (Ithaca, New York: Cornell University Press, 1957), pp. 168-173. 282 6.3. Behaviorism Perhaps the best that can be done in the way of providing a more positive reason for being dubious about the claim that wants and beliefs are mental episodes or events is to attempt to show that they are, strictly speaking, superfluous so far as explaining human actions goes. But Goldman has exerted a great deal of effort in showing that the explanations offered of actions in the behavioral sci ences either already make use of concepts which are very similar to our everyday concepts of wanting and believing, or cannot be construed as rivals of our common-sensical ex planations of actions in terms of wants and beliefs. I shall not discuss the claim that many explanations of actions in the behavioral sciences make use of concepts which are very similar to our common-sensical notions of wants and beliefs because I believe that it is eminently clear that they do. Terms like "motive," "need," "goal," "attraction" are often used as stand-ins for "want," and terms like "cognition" and "expectancy" are frequently used to mean what is ordin arily meant by "belief."216 any case, the fact that explanations in the behavioral sciences frequently employ 216 See Goldman's discussion of the theories of Edward C. Tolraan and Kurt Lewin. Alvin Goldman, A Theory of Human Action, op. cit., pp. 131-136. 283 concepts which are very similar in their behavior to the use that is made of "want" and "belief" in our everyday explan ations of actions is irrelevant to the claim that wants and beliefs construed as mental events are superfluous in explanations of human actions. However, in contrast to those psychologists who are not adverse to using such purposive notions like wanting and believing, there are other psychologists who avoid the use of purposive notions and attempt to account for behavior solely in terms of "stimulus," "drive," "conditioning," etc. Perhaps the most prominent contemporary psychologist who 217 holds this position is Burrhus F. Skinner. Skinner attempts to accout for the vast majority of behavior in terms of "operant conditioning." Operant conditioning refers to the various processes by means of which various behaviors can be "strengthened" by reinforcing events. The successes or failures of operant conditioning is explained in terms of purely external physical events or factors such as•the histories of the organism's reinforcements, the periods of picked Skinner over numerous other "stimulus- response" theorists because he seems to be the contemporary psychologist who is most averse to such purposive concepts as wanting and believing. 284 deprivation, and the stimuli presented to the organism. None of these factors contain the least hint of purpose or inten tion. This is, of course, only a very sketchy account of Skinner's system, but it will suffice to make my main point, viz., that Goldman's criticisms of Skinner's system do not help his case for the supposed mental status of wants and beliefs. So far as I can determine, Goldman's criticisms of Skinner's psychology are these: (1) there is nothing in Skinner's system which forbids the possibility that "Skinnerian" events first cause wants and beliefs which in turn cause behavior, and (2) if the follower of Skinner agrees that operant conditioning causes wants and beliefs, he cannot deny that these wants and beliefs also cause the behavior. Unfortunatley, the first claim is substantially unargued for by Goldman. The best that he has to offer in this regard is simply the fact that the discussions of other behaviorists seem to corroborate a possible relationship between "Skinnerian" events and desires and beliefs. 218 218 see Goldman's discussion of Miller and Dollard’s experiment. Alvin Goldman, A Theory of Human Action, op. cit., pp. 151-153. 285 Goldman's evidence for his second criticism is of two kinds: evidence drawn from "common sense," and evidence drawn from case studies under experimental conditions. First, the common-sense evidence: Here Goldman argues, Let us consider it a purely empirical matter, not to be decided by our definition of "act" or of "want," whether wants and beliefs are causally relevant to behavior. Let us assume that there are items of behavior such as taking a book, humming a tune, uttering a sentence, such that it is logically possible that they are not caused by wants and beliefs. Then let us ask whether, as a matter of fact, they are causally dependent on wants and beliefs or whether they are dependent on Skinnerian events alone, while the occurrence of wants and beliefs is quite incidental. As soon as the question is posed, however, the answer seems to me clear. Who can seriously deny that a person would act one way if he had certain wants and another way if he had other wants? Who can deny that my wanting to hum "Three Blind Mice" is causally relevant to my humming "Three Blind Mice," that if X had wanted to hum a different tune then I would have hummed a different tune? Similarly, who can deny that my beliefs are causally rele vant to my behavior? What could be more evident than that my belief that there is a lake north of town was a genuine determinant of my behavior of driving north on this hot day? Isn't it per fectly evident that if I had not believed there is a lake to the north, I would have driven in a different direction? 219 Surely, Goldman's argument here is simply trading on our ordinary understanding of the concept of a want, v/hereby the 219 Ibid., p. 154. 286 very notion of a want carries with it the idea of tending to cause the person who has it to act in certain ways. So, in a sense, as soon as the defender of Skinnerian events as the direct causes of human behavior admits the possibility that Skinnerian events cause wants and beliefs in addition to the behaviors, he cannot very well deny that wants and beliefs also cause behaviors. Indeed, there is in fact no earthly reason why he should insist on denying their causal significance once he has granted that they exist. Obviously, the crucial step is in admitting wants and beliefs into the picture in the first place. The "purely empirical matter1' is not whether wants and beliefs cause behavior, but whether there are wants and beliefs. The defender of Skinnerian events'strategy here would be to argue that the two behaviors mentioned by Goldman, viz., humming "Three Blind Mice" and driving north on a hot day, could be accounted for solely in terms of Skinnerian events. Other people are better equipped to do this than I am, but it is not difficult in the least to imagine how driving north on a hot day may be explained in terms of Skinnerian events. An inviting phone call from a voluptuous female to a playboy may cause him to drive north on this particular hot day, given a past history of conditioning to pleasure-seeking and a certain state of deprivation. 287 Of course, this rather simplistic account does not explain why he drove rather than walked or flew or why he drove north rather than in some other direction. But it is by no means clear that these specific behavioral responses could not be accounted for in terms of the history of reinforcements, states of deprivation of the individual subject, and various stimuli without ever mentioning wants and beliefs. Thus, Goldman has not demonstrated that as a matter of fact, wants and beliefs are necessary in any purely causal account of human behavior. In spite of all this, it may be claimed that this "common sensical evidence" mentioned by Goldman is strong enough such that if one desires to account for our common-sensical way of explaining actions, he could do better than identify or correlate wants and beliefs with "Skinnerian" events. It would seem that a much more plausible candidate for wants and beliefs is neurophysiological events and structures. This possibility will be examined in the next section. The second kind of evidence Goldman offers is two case studies under experimental conditions. The first of these is the Miller-Dollard example of a little girl who is known to be hungry and like candy; she is brought into a room where some candy has been hidden under a book in the bookcase. She is told this and that she may eat the candy 288 if she finds it. It takes the girl 210 seconds to find the candy on the first trial, 86 seconds on the second trial, 11 seconds on the third trial, and 86 seconds on the fourth trial. Yet, the history of her conditioning during the trial suggests that the time involved in her fourth trial would be an improvement over the third, but it isn't. Miller and Dollard try to account for this by postulating some intervening uncontrolled factor such as the little girl saying to herself, "This is too easy, he'll probably change the hiding place now that I know it." Now Goldman takes this to mean that the girl's beliefs about the hiding place were different than was expected and hence accounted for the difference in the expected behavior. But, as Goldman himself admits, this does not show that the girl's behavior cannot be accounted for in stimulus-response terms. And although the example does not show that the girl does not have beliefs and wants which are causally relevant in producing her behavior, it also does not show that she does have beliefs and wants and that these are causally relevant in producing her behavior. The second piece of experimental evidence concerns experiments which apparently show that wants and beliefs do not cause behavior. For example, the examiner asks the 289 subject to utter a sequence of single words, whatever words come to his mind. As the subject does so the examiner 11 reinforces'* any utterance of a plural noun by saying "Inmm“hImn,' approvingly, and, as could be expected, the fre quency of plural nouns increases, although the subject does not realize (believe) that this is happening. But again, rather than this showing that the subject does indeed have certain beliefs and wants which causally affect his behavior, all it does is show that the postulation of such wants and beliefs is consistent with the actual behavior. And we have already noted the ambiguity of be havior in this regard. The difficulty with Skinner's system and our ordinary criteria for belief and want ascriptions as well, it may be claimed, is that both of them deal with categories of behavior which are simply too large to be of much use in providing us with sophisticated tools we need for accurate and precise prediction and control. Both of these systems of explanation are analogous to an engineer's attempts to explain and duplicate the workings of a sophisticated computer simply by examining its input, output, and its history of programming. To dismiss in such a cavalier fashion the possible relevance of the internal workings of the human body may be pragmatically 290 justified for the moment, due to the impoverished state of our knowledge about this. But to dismiss it, proclaiming that everything that needs to be done can be done on the level of molar behavior, is hardly justified. Hence, I turn next to the possibility of identifying or at least correlating wants and beliefs with neurophysiological events and structures. 6.4. Neurophysiological Causes as Rival Explanations ' Of Human Actions One of the seemingly more plausible candidates for rival explanations of human actions is explanation in terms of neurophysiological causes. According to this view, all human bodily movements, including human actions, can be completely accounted for in terms of the electrical, chemical, and mechanical processes or states which occur in the body. Such accounts are, of course, not possible at the present moment, but the proponents of this view have claimed that there is no reason to suppose that this will always be the case. Some theorists, however, have interpreted the possibility of the eventual success of neurophysiological explanations of human actions as entailing the subsequent rejection of explanations in terms of our ordinary concepts of wants and beliefs. Hence, Norman Malcolm has written: 291 It has often been noted that to say B causes C does not mean merely that whenever B occurs, C occurs. Causation also has subjunctive and counterfactual implications: if B were to occur, C would occur; and if B had not occurred, C would not have occurred. But the neurophysiological theory would provide sufficient causal conditions for every human movement, and so there would be no cases at all in which a certain movement would not have occurred if the person had not had this desire or intention. Since the counterfactual would be false in all cases, desires and intentions would not be causes of human movements. They would not even be sufficient causal conditions, nor would they ever be necessary causal conditions. 220 As Goldman points out, Malcolm's claim that desires and intentions could never be sufficient conditions for human movements is unargued for. However, the second claim, viz., that if neurophysiological states are sufficient for human movements, then desires and intentions are not necessary for human bodily movements, seems to follow directly from the very definitions of the terms "sufficient’' and "necessary" themselves. That is, to assert that A is sufficient for B seems to be the same thing as saying that nothing other than A is required for B, i.e., that nothing over and above A is necessary for B. In other words, Malcolm's argument pre supposes the following principle,: (I) If events C^, . . . Cn are jointly sufficient for the occurrence of event E, then no events other than C^, . . . Cn are necessary for the occurrence of E. 220cited by Goldman, Ibid., p. 158. 292 Goldman, however, challenges this principle on the grounds that it rules out the possibility of chains of causes, and he presents the following argument which he believes shows that causal chains are ruled out by the principle: . . . suppose that there are events C^, . . . Cn which occur simultaneously at t2 and which together with laws of nature, are sufficient for the occur rence of E at tjj. Further suppose that C]_, . . . Cn are simply one group of events in a sequence of causes which are sufficient for the occurrence of C-j_, . . . Cn at tj, that C^, . . . C are suf ficient for certain events at t3, and that these events at t3 are sufficient for E at t4# Now since Clf . . . Cn are necessary for the occurrence of E, principle (I) implies that no events other than C-^, . . . Cn are necessary for the occurrence of E. From this it follows that no events either be fore or after t2 are necessary for the occurrence of E. This means that the indicated events at t^ or t^ are not necessary for the occurrence of E. But if they are not necessary for the occurrence of E, then they are not a cause of E. This, however, contradicts our original assumption - viz., that each of these groups of events is a cause of E. 221 This argument appears plausible enough, but I believe that it can be shown not to be so. The plausibility of the argument clearly depends on an ambiguity in the phrase "cause of an event." In one sense of this phrase it refers to those conditions without which the event in question would not have occurred. This is the sense in which the 221 ibid., p. 160. 293 phrase is said to denote a necessary condition for the occurrence of the event in question, and the sense in which it is appropriate to speak of a cause of an event. The second and quite distinct sense in which the phrase "cause of an event" is used is the sense in which it refers to that condition or set of conditions whose presence will result in the occurrence of the event in question. This is the sense in which "cause of an event" refers to the sufficient condition for the occurrence of an event, and the sense in which it is appropriate to speak of the cause of an event as opposed to a cause of an event. The above argument, however, begins by assuming that what it means to say that there are chains of causes is that there is a sequence of events, all of which precede event E and are necessary for the occurrence of E. That is, the argument clearly pre supposes that the events occurring at time t^, call them B-^, . . . Bn ; the events occurring at time t2, call them Ci, . . . Cn ; and the events occurring at time t2, call them , . . . Dn are all equally causes, i.e., necessary conditions for the occurrence of event E at time t^. Yet, the senses in which we may appropriately speak of a chain of causes is not restricted to this sense only. For we may appropriately speak of a chain of causes whenever the events in a sequence of time intervals each considered separately 294 constitutes a sufficient condition for the occurrence of the events which occur in the immediately succeeding time interval. Hence, B ^ . . . Bn at tj^? C1; . . . Cn at t2; D^, . . , Dn at tj? E at t4 may be considered to be a chain of causes since B^, . . . Bn is a sufficient condition for the occurrence of C^, . . . Cn at t2? C^, . . . Cn at t2 is a sufficient condition for the occurrence of . . . Dn at t3; and D^, . . . Dn at t3 is a sufficient condition for the occurrence of E at t4. Now, the argument claims that B1# . . . Bn; Ci, . . . Cn; , . . . Dn are all causes of E at t4 in the sense in which each one of these is a neces sary condition for the occurrence of E at t4 , but it fal laciously identifies this claim with the quite distinct claim that B^, . . . Bn is sufficient for C^, . . . cn; Clt . . . Cn is sufficient for DL, . . . Dn ; and Dlf . . .d is sufficient for'E. A different way of seeing the flaw in this argument is to notice that it presupposes the truth of the doctrine of the plurality of causes and effects. For the argument asks us to suppose that, by virtue of certain laws of nature, events C^, . . . cn at t2 are sufficient for the occurrence of E at t^; the events C^, . . . Cn at t2 are sufficient for D^, . . . Dn at t2; and the events D^, . . . Dn at t3 are sufficient for the occurrence of E at t4. But if both 295 C-j_, . . . cn at t2 and . • • Dn at t3 are causally sufficient for the occurrence of E at t^, then there are at least two causes of E, and not just two causes in the sense of there being two different sets of sufficient conditions for E. Likewise, C^, . . . Cn at t2 is causally sufficient for both D]_, . . . Dn at t3 and E at . Now, although it seems apparent that a plurality of causes often occurs, it is by no means clear that we are justified in assuming that plurality of causes is a feature of the real world rather than merely a feature of our limited ability to understand the world. The same effect, it would appear, may occur under a number of different sets of con ditions. If you want to remove a spot from a garment, for instance, you can do so by using gasoline, or lighter fluid, or carbon tetrachloride, or a large variety of other chemical agents. You can get expelled from school by blowing up a building, by writing slanderous statements in the school newspaper, and so on. Yet, in spite of the intuitive ob viousness of such claims, it may be argued that a large number, and perhaps all, such plurality of causes are only apparent and for two reasons. First, many apparent cases of plurality of causes result from including too much in the statement of the sufficient conditions or cause. Hence, it may be claimed that a certain billiard ball A v/as caused to 296 move in a certain direction and with a certain speed not merely because it was struck by another billiard ball, but by being struck with a cue stick, or by being pushed by one's hand, or by placing i,t on an uneven table. But is this unequivocally a genuine case of plurality of causes? I think not. What is sufficient (and also necessary) for the ball to move in this direction and at that speed is that a certain amount of force be applied to it in that direction. It does not in the least matter exactly what is responsible for the force being applied. Another billiard ball, the cue stick, one's hand, and the gravitational attraction of the earth are all equally irrelevant to enumeration of the precise condition(s) upon which this event depends. The second reason for rejecting these cases of plurality of causes as only apparent is that frequently the effect event is characterized too broadly. Hence, it appears as if a house can burn down as a result of defective electrical wiring, lightning, a defective furnace, and so on. Yet a trained fireman can frequently tell from an examination of the ruins which one of these conditions was the actual cause of the fire. "House fire" is at most only a very general description of the effect, and rather than showing the plurality of causes it merely demonstrates the need for a 297 more precise breaking down of effects, and perhaps if causes and effects were broken down in a more careful manner, there would be much less plausibility in the doctrine of plurality of sufficient conditions. Not everyone, however, has been prepared to abandon the doctrine of the plurality of sufficient causes in view of the foregoing arguments. Arthur Pap, for ex ample, has claimed that the foregoing arguments are exceedingly weak, and for two major reasons.222 The first reason is that there are some highly specific effects which can be shown to have a plurality of sufficient causes. The second reason is that the requirement that causes and effects be described in complete detail, which seems to be the guiding principle in the fore going arguments, would result in the destruction of all causal laws. The first example which supposedly shows the plurality of logically independent sufficient condi tions is borrowed from mechanics. For, according to Newton's second law of motion, the direction and magnitude of every accelerated motion is uniquely determined by a force acting in the same direction. This force can, how ever, be resolved into component forces, according to the 222 Arthur Pap, An Introduction to the Philosophy of Science (New York: The Free Press of Glencoe, 1962), pp. 255-258. 298 law of vector addition. Thus, if we conceive of two horses of unequal strength pulling a cart simultaneously in op posite directions, the resulting acceleration of the cart would be uniquely determined by the difference in the two opposite forces. But if these opposing forces are doubled (conceive of matching each horse with one of equal strength), then exactly the same acceleration of the cart would result, but in each case it is produced by a conjunction of dif ferent component forces. Thus, we have a case of the same effect produced by two different sets of sufficient con ditions . The second counterexample presented by Pap concerns the straight movement of a billiard ball on a billiard table. Now, if we identify the effect as "the billiard ball's passage from point p to point p' in t seconds," then a physicist could predict this effect if he had known the exact impulse the ball received and the exact frictional resistance of the table top. But he could also determine the exact amount of impulse and direction which would have resulted in the same effect even if the passage from point p to point p' involved the billiard ball's bouncing off the table edge or another ball. Thus, there is in this case more than one sufficient condition for the effect. 299 But no matter how initially plausible these counter examples are, they certainly may be challenged. For it is apparent that the plausibility of these counterexamples depends upon Pap's analysis of singular causal statements. On Pap's view, a singular causal statement of the form "a caused (will cause) b," where a and b represent particular events, is to be analyzed as follows: A and B denote re peatable kinds of events of which a and b are instances; then "a caused (will cause) b" means that a was (will be) followed by b in the presence of C, and an instance of A is followed by b in the presence of C, and an instance of A is followed by an instance of B whenever C is present, where C is the sum total of conditions which is necessary for A to be followed by B.223 The necessity of including C here springs from the fact that most singular causal statements, e.g., "My stepping on the gas pedal caused the car to start moving," fail to make mention of the fact that some ad ditional conditions C are presupposed. Obviously, my step ping on the gas pedal will not result in the car's moving unless certain other conditions are present, e.g., there is gasoline in the car, the motor is running, the drive shaft is connected, and so on. In other words, there is no such thing as an unconditional cause. 223 ibid., p. 255. 300 Yet it can be freely admitted that this is the sense in which "cause" is most frequently used in singular causal statements without at the same time committing oneself to the supposition that what is referred to as the cause in such cases is the cause in any real sense. For no one who is the least bit knowledgeable about the world would suppose that my stepping on the gas pedal is the cause of the car starting to move, even though we frequently say it is the cause. The most correct thing to say in this instance is rather that my stepping on the gas pedal together with these other conditions causes the car to start moving. In short, all that Pap has shown is that what we frequently mention as the cause of a particular event is something less than a sufficient condition; he has not shown that it is in any sense impossible to mention all the conditions which together constitute a sufficient condition for the occurrence of a particular event. As regards the first of Pap's two counterexamples, it may be claimed that it is not true that compounding forces in the manner prescribed in this example will inevitably result in identical effects. For clearly, if the cart is of flimsy construction, the compounding of forces will not result in an equal acceleration of the cart. For in that case, the cart will be split in two. Likewise, one would 301 not expect similar results when one first applies opposite but unequal pressure to an uncooked egg with his fingers as when he merely continues to increase these opposite forces equally. The fault in the second counterexample is the topic of Pap's second attempt to defend a plurality of causes. That is, the example, it is claimed, may be dis missed as only apparent since it is a case in which the effect, i.e., the billiard ball's passage from point p to point p' in t seconds, has been described too broadly. Against the argument that the plurality of causes is merely an appearance arising from insufficiently specific descriptions of the effect, Pap argues, But all this is irrelevant, since any causal statement, by the very meaning of "cause" and "effect," involves abstractions of recurrent features from the concrete, individual event. That e^ is an instance of the more specific kind of event EA, and &2 an instance of the more specific kind of event EB, is surely compatible with saying that both are instances of the less specific kind of event E . . . If, on the other hand, an effect were, per impossibile, described in complete detail, its sufficient condition would be unique only in the trivial sense that the effect itself would be unrepeatable. 224 Pap's point here is, I believe, a sound one, and one which has far-reaching implications for the thesis that neurophysiological explanations are in competition with our 224ibid., pp. 257-258. 302 common-sensical want and beliefs explanations. The point is a point about explanation. For, if what I have argued in Chapter III is correct, then to explain the occurrence of an event b by referring to another event a, is to imply that there is some lav/ or law-like statement to the effect that events of kind B are invariably connected with events of kind A. Hence, if citing the occurrence of event a is going to succeed in explaining the occurrence of event b, a and b must be viewed as particular events of certain kinds of which there may be further instances. To claim that the individual event b can be explained as caused by another individual event a does not mean that whenever a occurs b occurs, since a and b are individual events which have particular spatiotemporal locations and hence, occur only once. Thus, it seems that we must leave open the pos sibility that there can be a plurality of sufficient causes and, with it, even the possibility of mind-body interaction- ism. The possibility of such neurophysiological explanations envisioned by Malcolm and others, then, does not preclude ex planations in terms of wants and beliefs. Although Goldman does not discuss the doctrine of the plurality of sufficient causes, he apparently suspects that his argument against principle (I) has failed to show that neurophysiological explanations are not rivals of our 303 common-sensical explanations. What he recognizes as the flaw in his argument is that it assumes that some events occurring at some other time (either before or after) than the time at which the events which are sufficient for the occurrence of E may be necessary for the occurrence of E. However, the neurophysiological activity in question may be continuous activity leading eventually to the occurrence of the overt behavior. In this case, if wants were to cause the behavior, they would have to be simultaneous with the set of neurophysiological events that are causally suf ficient to produce the behavior. And in view of this, Goldman attributes the following weaker principle to Malcolm's argument: (II) If events C* occurring at t, are sufficient for the occurrence of E at t2 , then no other events at t^ are necessary for the occurrence of E at t2 . 225 Goldman, however, thinks that this principle can be proven to be false. His argument is as follows: Suppose there is a law saying that for any object o and any time t, the object has property $ at t if and only if it has property V at t. Then if a particular object o has properties $ and V at a particular time t^, I shall say that o's having $ at t^ is a "simultaneous nomic equivalent” of o's having V at t]_. Now suppose that particular events 225 Alvin Goldman, A Theory of Human Action, op. cit., p. 161. 304 (or sets of events) C* and C** are simultaneous nomic equivalents. This means that C* is suf ficient for.C** and that C** is sufficient for C*; equivalently, it means that C* is necessary for C** and that C** is necessary for C*. Further, suppose that C* is sufficient for the occurrence of a subsequent E (at t2) and necessary for the occurrence of E. Then there is another event (or set of events) - viz., C** - that is simul taneous with C* and is also necessary for the occurrence of E. 226 The difficulty with this argument is that it may plausibly be argued that there are no such laws relating neural and mental events, and there would have to be if Goldman is correct in claiming that occurrent wants and beliefs are mental events. Evidence for the truth of this claim can be gathered from the extreme difficulty Goldman exper iences in attempting to formulate a law which would connect wants and beliefs with acts. In formulating the notions of act generation, basic act-type, basic act-token, and intentional action, it was tacitly assumed that the agent had-only one set of relevant beliefs and desires. Yet, since human beings frequently have many beliefs and desires which are relevant, some way of relating these with the act that the agent performs must be formulated. The best that 226jbid., p. 161. 305 Goldman can come up with is the following law: If any agent S believes that hypothetical act-tree Al' a 2' * * * An ae Performet^ at is more likely, all in all, to achieve more of his desires than any other act-tree (that could be performed at t), and if S is in standard conditions with respect to each of the basic acts of this act-tree (at t), then S performs each of these basic acts (at t) . 227 This "law," however, is extremely vague, as Goldman himself admits. Its vagueness can be clearly seen if we consider two hypothetical act-trees, A and B. Act-tree A has a higher probability of satisfying lesser desires and act- tree B has a smaller probability of satisfying greater desires. Which one will agent S choose? About all that can be said here is that S 1s choice will depend upon what ever decision procedure S uses in making his choice. But this is highly uninformative since we might suppose that this decision procedure varies from person to person. This problem is further complicated by the fact that, at present, we lack adequate techniques for measuring the relative strengths of desires. Solutions to these problems seem to be in the offing if occurrent wants and beliefs can be identified with certain neural events,and dispositional wants and beliefs can be 227 ibid., p. 74. 306 identified with certain neurological states or structures. Just such an identity has been put forward by Herbert Feigl228 and J.J.C. Smart. 22® Such a theory, unlike analyti cal behaviorism, is attractive simply because it does not require a behavioral analysis of all belief and want state ments. For on this theory, although belief and want state ments do not have the same meanings as statements about neurological events, states, and structures, they all do refer to the same things. This theory seems to be quite plausible even if no physiologist has been successful in finding a one-to-one correspondence of neurological events, states, and structures with specific wants and beliefs. Such a successful correlation would constitute a large step forward in providing answers to our remaining problems of determining what an individual's decision procedure is and measuring the relative strengths of desires. Thus, for example, an individual's decision "criterion" may be cor related with the complex structure of the connections in the 22® Herbert Reigl, "Mind-Body, Not a Pseudoproblem," in Dimensions of Mind, ed. by Sidney Hook (New York: Collier Books, 1961), p. 38. 229 J.J.C. Smart, "Sensations and Brain Processes," in The Philosophical Review, LXVIII (1959), pp. 144-145. 307 cortex and the relative intensities of desire may be successfully correlated with the relative number of such cell assemblies involved in a given situation. It may be argued in response to this that since, for the present, we have more information about why people behave the way they do from our common-sensical observations and psychology than we have from neurophysiology, human actions can be more fruitfully explained in terms of wants, beliefs, etc., than in terms of neurological events, states, and structures. But this objection is at most merely a pragmatic one, and does nothing to show that there is any intrinsic implausibility in the view that wants and beliefs will eventually be found to be identical with these neuro logical events, states and structures. CHAPTER VII CONCLUSIONS At this point I believe that I may regard the task which I set at the beginning of this dissertation as com pleted. The task which I outlined at the beginning of this dissertation was to find answers to the two questions: (1) To what extent is it possible to discover clear and precise criteria for distinguishing human actions from mere physical movements? and (2) To what extent is the claim that reasons are not causes justified? With regard to the first question, we have seen from our examination of the failure of a number of philosophers' attempts at analyzing the concept, that it appears to be an essentially contested concept. This is expecially true if there are mental acts as well as overt acts. But we have also seen that a large part of the reason for this failure springs from a failure to see that there are at least two contrasts marked by the term "action." For if we restrict our investigation to overt acts, then these can be divided into two major categories: bodily acts and the acts which are either directly or indirectly generated from them. But since it was discovered that the various kinds of act- generation each involve talking about circumstances and that 308 309 there was no way in which these circumstances could be completely exhausted, it was concluded that with regard to some of these generated acts no totally satisfactory analysis of the form ’’Whenever x, y and z, S signals" can be provided. The second class of actions, bodily actions, could be analyzed in terms of the specific kind of causation which brought them about. The second major question, viz., "To what extent is the claim that reasons are not causes justified?" was also examined, and it was found that none of the argu ments proposed by the action theorists for this con clusion were justified. An attempt was also made to elucidate what action theorists like Melden, Strawson, Anscombe, and Louch think is the sense in which citing an agent's reasons succeeds in explaining his actions. However, what was found in every case was that these reason, concepts were claimed to be evaluative principles of some sort which involved the idea of a conscious person. But it was shown that not only is it not clear in what sense the concepts of reasons, persons, and human actions are evaluative, but it was also argued that ex planations of a person's actions in terms of his reasons for acting presuppose general truths or laws about human beings. There is then no irreducible gap between rational explanations and scientific explanations, and rational 310 explanations do not differ radically from scientific explanation. There is no good reason then to assume that human actions are to be understood in ways which are methodologically significantly different from the ways in which the rest of nature is understood. One may then significantly search for neurophysiological corre lates or identities for wants and beliefs. If what I have been arguing is correct, then it con stitutes a means by which this dispute between some philosophers and some psychologists may be settled. For the philosophers who have claimed that reason explanations are of a radically different sort from scientific explana tions because reason statements do not refer to events are wrong because reason statements frequently do refer to events. And likewise, the psychologists who claim that reason explanations are useless are wrong since ordinary discourse possesses a considerable number of distinctions which may prove useful to psychological theory. Further research may show that some revision in these conceptural distinctions may be necessary, but the psychologist would do well not to dismiss them all out of hand without further inquiry. I feel obligated to mention the philosophical sig nificance of this investigation into the concept of human action to the two problems that I mentioned in the beginning 311 of this dissertation: viz., its relevance to ethical debates and its relevance to the problem of determinism and freedom. Our discussion of the "accordion effect" whereby an act may be described narrowly so that it ex cludes consequences or broadly so that it includes con sequences suggests that the debate between the Ethical Formalist and the Ethical Teleologist is resolvable. For the Formalist claims that the rightness, wrongness, or obligatoriness of an act is a function solely of what kind of act it is without regard to its consequences, while the Teleologist argues that the rightness, wrong ness, or obligatoriness of an act is a function solely of its good or bad consequences. But if we are free, as the "accordion effect" seems to suggest, either to describe an agent's act so that the description includes the consequences or we describe the act and its conse quences separately, then there is nothing to choose between insofar as these two theories are concerned. And if there is nothing to choose between insofar as these two theories are concerned, they are not rival theories. With regard to the philosophical significance of our investigation into the concept of human action for the problem of determinism and freedom, it would appear that those philosophers who appeal to the concept of a human action as showing that at least some human behaviors are uncaused are mistaken at least about overt acts. For, if what I have been arguing is correct, all such acts require some sort of a bodily action be performed by the agent, and the distinguishing feature of bodily acts is that they are caused either by the agent's wants or by his wants and beliefs. And although I have not dis cussed the question of whether or not an agent's wants and beliefs are caused, I know of no evidence to the contrary. And since there is a good possibility that wants and beliefs may be identical with neurophysiological events, states, and/or structures, and everyone believes these are caused, there is no absurdity in the notion that wants and beliefs are caused. Hence, I do not be lieve that the concept of human action constitutes any thing like a secure citadel in which human freedom might defend itself against determinism. Finally, I feel obligated to mention some of the problems that I have only mentioned in this dissertation without attempting to provide any sort of solution to them. The most obvious one is the question of the possi bility of the existence of mental acts and events. But as I have repeatedly claimed, I simply do not know of any way in which we could either prove or disprove that such things exist. The other problem is the problem of determining what decision criteria an individual uses, and using it to predict and explain his actions. This is a problem, however, which is best left to someone who is competent in the theory of decision making. BIBLIOGRAPHY Abelson, Raziel. "Cause and Reason in History." Philosophy and History. Edited by Sidney Hook. New YorJci New York' University Press, 19 63. Abelson, C. Raziel. "Doing, Causing, and Causing to Do." The Journal of Philosophy, LXVI (19 69). pp. 178-192. Alston, William P. "Motives and Motivation." The Encyclo pedia of Philosophy. Vol. V. Anscombe, G. E. M. Intention. Ithaca, New York: Cornell University Press^ 1969. Anscombe, G. E. M. "Intention." The Philosophy of Action. Edited by. Alan R. White. London: Oxford University Press, 1968. Aristotle. Nicomachean Ethics. The Basic Works of Aristotle. Edited by Richard McKeon. Translated by W. D. Ross. New York: Random House, 1941. Austin, J. L. How to Do Things with Words. Cambridge, Massachusetts: Harvard University Press, 1962. Austin, J. L. "Ifs and Cans." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 19 70. Ayer, A. J. Man As A Subject for Science. Readings in Ethical Theory. Edited by Wilfred Sellars and John Hospers. 2nd Edition. New York: Appleton-Century- Crofts, 1970. Baier, Kurt. "Action and Agent." Monist, XLIX (1964), 183-195. Baier, Kurt. "Responsibility and Action." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 19 70. Benn, S. I., and Peters, R. S. "Freedom and Responsibility." Social Principles and the Democratic State. London: George^ Allen and Unwin, Ltd., 1959. Benson, John. "The Characterization of Actions and The Virtuous Agent." Proceedings of the Aristotelian Society (New Series), LXIII (1962-3). 314 315 Bousma, ,0. K. "Moore's Theory of Sense-data." The Philoso phy of G. E. Moore. Edited by P. A. Schi1pp. Evanston, Illinois: Northwestern University Press, 1942. Braithwaite, R. B. Scientific Explanation. Cambridge: Cambridge University Press, 19 64. Brand, Myles. "Danto's Basic Actions." Nous, II (1968) pp. 187-90. Brand, Myles, ed. The Nature of Human Action. Glenview, Illinois: Scott, Foresman and Company, 1970-. Brodbeck, May. "Meaning and Action." Philosophy of Science, XIX {October, 1963). 1 Chisholm, R. M. Perceiving. New York: Cornell University Press, 1957. Chisholm, Roderick M. Theory of Knowledge. Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966. Copi, Irving. Introduction to Logic. 2nd Edition. New York: The Macmillan Company, T961. Danto, Arthur. "Basic Action." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 1970. Danto, Arthur. "What We Can Do." The Journal of Philosophy, LX (1963), p. 436. D'Arcy, Eric. Human Acts: An Essay in Their Moral Evalua tion. Oxford, England: Oxford University Press, 1963. Davidson, Donald. "Actions, Reasons, and Causes." The Nature of Human Action. Edited by Myles Brand. Glen- view, Illinois: Scott, Foresman and Company, 1970. Davidson, Donald. "The Logical Form of Action Sentences." The Logic of Decision and Action. Edited by Nicholas Reseller. Pittsburgh: University of Pittsburgh Press, 1967. Dray, William. Laws and Explanation in History. London: Oxford University Press, ,1957. Dray, William. Laws and Explanation in History, Chapter V, Section 4. Readings in the Theory of Action. Edited by Norman S. Care and Charles Landesman. Bloomington, Indiana: Indiana University Press, 1968. 316 Dray, William. "The Historical Explanation of Action Reconsidered." Philosophy and History. Edited by Sidney Hood. New York: New York University Press, 1963. p. 86. Ewing, A. C. "May Can-Statements Be Analysed Determin- istically?" Proceedings of the Aristotelian Society, LXIV (1964), pp. 171-2. Peigle, Herbert. "Mind-Body, Not a Pseudoproblem." Dimensions of Mind. Edited by Sidney Hook, New York: Collier Books, 1961. Feinberg, Joel. "Action and Responsibility." The Philosophy of Action. Edited by Alan R. White. London* Oxford University Press, 1968. Geach, P. T. "Ascriptivism." Ethics. Edited by Judith J. Thompson and Gerald Dworkin. New York: Harper and Row, 19 68. Gean, W. D. "Reasons and Causes." Review of Metaphysics, XXX (June, 1966). Goldman, Alvin I. A Theory of Human Action. Englewood Cliffs, New Jersey: Pr ent'ice-Ha 11V Iric., 1970. Goodman, Nelson. "The Way the World Is." Review of Meta physics^ XIV (1960), pp. 48-56. Hampshire, Stuart, Freedom of the Individual. London, England: Chatto and Wendus, Ltd., 196 5. Hare, R. M. Freedom and Reason. London, England: Oxford University Press, 19 63. Hare, R. M. The Language of Morals. London, England: Oxford University Press, 19 52. Hart, H. L. A. "The Ascription of Responsibility and Rights." Logic and Language, First Series. Edited by Antony Flew. Oxford: Basil Blackwell, 1951. Hart, H. L. and A. M. Honore. Causation in the Law. Oxford: Oxford University Press, 1957. Hempel, Carl G. Aspects of Scientific Explanation. New York: The Free Press, 1965. 317 Hempel, Carl G. "Explanatnion in Science and in History." An Introduction to Philosophical Inquiry. Edited by Joseph Margolis. New York: Alfred A. Knopf, 1968. Hempel, Carl G. Philosophy of Natural Science. Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966. Hempel, Carl G. "Reasons and Covering Laws in Historical Explanation." Philosophy and History. Edited by Sidney Hook. New York: New York University Press, 1963. Hempel, Carl G. "The Function of General Laws in History." Reading in Philosophical Analysis. Edited by Herman Peigl and Wilfrid Sellars. New York: Appleton- Century-Crofts, Inc., 1949. Hempel, Carl G. and Oppenheim, Paul. "Studies in the Logic of Explanation." Philosophy of Science, XV (1948), p. 137. Hempel, Carl and Oppenheim, Paul. "The Logic of Explanation." Readings in the Philosophy of Science. Edited by H. Feigl and M. Brodbeck. New York: Appleton- Century-Crofts, Inc., 19 53. Hospers, John. An Introduction to Philosophical Analysis. 2nd Editionl Englewood Cliffs, New Jersey: Prentice- Hall, Inc., 1967. Hospers, John. "What Is Explanation?" Essays in Conceptual Analysis. Edited by Antony Flew. New York: St. Martin's Press, Inc., 1956. Hume, David. A Treatise of Human Nature: Being An Attempt to Introduce the Experimental Method of Reasoning Into Moral Subjects. Edited by L. A. Selby-Bigge. Oxford, England: Oxford University Press, 1888. Hunt, Ivor. "Value-Concepts and Conceptual Truth." Proceedings of the Aristotelian Society. LXIII (1962): Lehrer, Keith. "An Empirical Disproof of Determinism?" Determinism, Free Will, and Moral Responsibility. Edited by Gerald Divorkin. Englewood Cliffs: Prentice-Hall, Inc., 1970. Lehrer, Keith. "Ifs, Cans and Causes." The Nature of Human Action. Edited by Myles Brandi Glenview, Illinois: Scott, Foresman and Company, 1970. 318 Louch, A. R. Explanation and Human Action. Berkeley and Los Angeles, California: University of California Press, 1966. Melden, A. I. "Action." Readings in the Theory of Action. Edited by Norman S . Care and Charles Landesman. Bloomington, Indiana: Indiana University Press, 19 68. Melden, A. I. Free Action. London: Routledge and Kegan. Paul, 1965: Melden, A. I. "Willing." Philosophical Review, LXIX (1960), pp. 475-84. McCall, Storrs. "Ability as a Species of Possibility." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 1970. Nagel, Ernest. The Structure of Science: Problems in the Logic of Scientific Explanation. New York: Harcourt, Brace and World, Inc., 1961. Oldenguist, Andrew. "Choosing, Deciding, and Doing." The Encyclopaedia of Philosophy. Vol. II. Pap, Arthur. An Introduction to the Philosophy of Science. New York: The Free Press of Glencoe, 19 62. Peters, R. S. The Concept of Motivation. 2nd ed. London: Routledge and Kegan Paul, 1960. Peters, R. S. The Concept of Motivation. New York: Humanities Press, i960: Pitcher, George. "Hart on Action and Responsibility." Philosophical Review, LXIX (1960), pp. 226-35. Prichard, H. A. "Acting, Willing, Desiring." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 1970. Putnam, Hilary. "Robots: Machines or Artificially Created Life?" Philosophy of Mind. Edited by Stuart Hamp shire. New York: fjarper and Row, 1966. Quine, Willard Van Orman. From a Logical Point of View. 2nd ed. New York: Harper and Row, Inc., 1963. Quine, Willard Van Orman. Word and Object. Cambridge, Massachusetts: The M. I. T, Press, 1960. 319 Rescher, Nicholas. "Aspects of Action." The Logic of Decision and Action. Edited by Nicholas Rescher. Pittsburgh, Pennsyivania: University of Pittsburgh Press, 1966. Rescher, Nicholas. "On the Characterization of Action." The Nature of Human Action. Edited by Myles Brand. Glenview, Illinois: Scott, Foresman and Company, 1970. Rudner, Richard S. Philosophy of Social Science. Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1966. Ryle, Gilbert. The Concept of Mind. New York: Barnes and Noble, 1949. ‘ Scheffler, Israel. The Anatomy of Inquiry.' New York: Alfred A. Knopf^ 19 63. Schlick, Moritz. Problems of Ethics, A Modern Intro duction to Philosophy. EdltedTiy^aul Edwards and Arthur Pap. New York: The Free Press of Glencoe, 1957. Scriven, Michael. "Explanations, Predictions, and Laws." Minnesota Studies in the Philosophy of Science, Vol. Ill: Scientific Explanation, Space and Time. Edited by Herbert Feigl and Grover Maxwell. Min- neapolis: University of Minnesota, 1962. Smart, J. J. C. "Sensations and Brain Processes." The Philosophical Review. LXVIII (1959), pp. 144-145. Stevenson, C. L. Ethics and Language. New Haven: Yale University Press, 1944. Strawson, P. R. Individuals. Garden City, New York: Anchor Books, Doubleday and Company, Inc., 19 63. Taylor, Richard. Action and Purpose. Englewood Cliffs, New Jersey: Prent1ce-Hall, Inc., 1966. Toulmin, S. E., and Baier, K. "On Describing." Philosophy and Ordinary Language. Edited by Charles E. Caton. Urbana, Illinois: University of Illinois Press, 1963. White, Alan R., ed. The Philosophy of Action. London: Oxford University Press, 19 68. 320 Wittgenstein, Ludwig. Philosophical Investigations. Translated by G. E. Anscombe. New York.: The Mac- millan Company, 1958. Wright, Georg Henrik Von. Norm and Action: A Logical Enquiry. New York: Routledge and Kegan Paul, I5"6 3. Ziff, Paul. "About Behaviorism." The Philosophy of Mind. Edited by V. C. Chappell. Englewood Clifts, New Jersey: Prentice-Hall, Inc., 1962.