You are here


The present work is a thoroughly revised version of the first of the two series of Gifford Lectures on ‘Norms and Values’, which I gave at the University of St. Andrews in 1959 and 1960. The content of the second series was published in 1963 in the International Library of Philosophy and Scientific Method under the title The Varieties of Goodness. The latter work and the present are substantially independent of one another. There is, however, a minor amount of overlap between the discussion of the ontological status of prescriptions in Chapter VII of this book and the discussion in the last three chapters of The Varieties of Goodness.

In 1951 I published in Mind a paper with the title ‘Deontic Logic’. In it I made a first attempt to apply certain techniques of modern logic to the analysis of normative concepts and discourse. Since then there has been a growing interest in the logic of norms among logicians and, so far as I can see, among legal and moral philosophers also. Moreover, the name deontic logic, originally suggested to me by Professor C. D. Broad, seems to have gained general acceptance.

The thoughts which are contained in the present work are the fruits, partly of criticism of ideas in my early paper, and partly of efforts to develop these ideas further. I should like to say a few words here about the growth of my thoughts and the plan of this book. Readers who have no previous familiarity with deontic logic may skip this part of the Preface.

In my original paper the two ‘deontic operators’, O for obligation and P for permission, were regarded as interdefinable. O was treated as an abbreviation for ~P~. The operators were prefixed to what I regarded as names of acts, A, B… and to molecular compounds of such names. The acts were conceived of as act-categories, such as, e.g., murder or theft, and not as act-individuals, such as, e.g., the murder of Caesar. Act-categories were treated as ‘proposition-like entities’, i.e., entities on which the truth-functional operations of negation, conjunction, disjunction, etc., can be performed. The meanings of expressions such as OA or P(A & ~B) I regarded as propositions to the effect that certain categories of acts are obligatory or permitted. Thus the possibility of combining the expressions by means of truth-connectives was taken for granted. I also, however, implicitly regarded these meanings as norms enjoining or permitting acts. It did not then occur to me that this made the applicability of truth-connectives to the expressions problematic. Since the expressions OA, etc., obviously could not themselves be regarded as names of acts, iterated use of the operators O and P was not allowed by the rules of the calculus. Expressions such as OOA were accordingly dismissed as meaningless.

I have since come to entertain doubts on practically all issues of importance in my first publication on deontic logic. These doubts have been of two kinds. Some concern the validity of certain logical principles of obligation-concepts, which I had originally accepted. Others concern the interpretation of the symbols and expressions of the calculus.

One of my doubts of the first kind relates to the nature of permissive norms. Is permission an independent normative concept, or can it be defined in terms of obligation (and negation)? If it can be so defined, what is the correct way of defining it? These questions are discussed in the last four sections (13–16) of Chapter V and briefly touched upon in various places elsewhere in the book.

Other doubts of the first kind have to do with the principles of distributivity of the deontic operators and the various principles of ‘commitment’. When these principles are formulated in a more refined logical symbolism it turns out that they do not possess the unrestricted validity which I originally claimed for them. These various laws of the ‘old system’ are discussed and corrected in the last four sections (15–18) of Chapter IX.

My dissatisfaction and doubts relating to questions of interpreting the calculus were even more serious, and became in the end destructive of the entire original system.

If A denotes an act, what does ~A mean? Does it signify the not-doing of the thing, the doing of which is symbolized by A? Or does it signify the undoing of that thing, i.e., the doing of something which results in an opposite state of affairs? If the first answer is the right one the question will arise what we are to understand by ‘not-doing’: the mere fact that a certain thing is not done, or the forbearance of some agent from doing this thing, when there is an opportunity to do it? If the second answer is correct, how shall we then distinguish between leaving something undone and undoing it?

These and similar considerations made it plain that the symbolism for acts which I had been using was inadequate for expressing logical features of action, which are of obvious relevance to a logic of obligation-concepts. The same inadequacy would have been there had I regarded A, B, etc., not as names of categories of acts, such as manslaughter or window-opening, but as sentences describing states of affairs, such as that a man is dead or a window open. In short, the symbolism of so-called propositional logic was inadequate for symbolizing the various modes of action. New logical tools had to be invented. A Logic of Action turned out to be a necessary requirement of a Logic of Norms or Deontic Logic.

We could say that formal logic, as we know it to-day, is essentially the logic of a static world. Its basic objects are possible states of affairs and their analysis by means of such categories as thing, property, and relation. There is no room for change in this world. Propositions are treated as definitely true or false—not as now true, now false. Things are viewed as having or lacking given properties and not as changing from, say, red to not-red.

Acts, however, are essentially connected with changes. A state which is not there may come into being as a result of human interference with the world; or a state which is there may be made to vanish. Action can also continue states of affairs which would otherwise disappear, or suppress states which would otherwise come into being. A necessary requirement of a Logic of Action is therefore a Logic of Change.

Our first step towards building a Deontic Logic will be to survey the traditional logical apparatus with a view to constructing out of its ingredients a new apparatus which is adequate for dealing, at least in gross outline, with the logical peculiarities of a world in change. This is done in Chapter II, which contains the fundamentals of a Logic of Change. After a general discussion of the concept of action in Chapter III the fundamentals of a Logic of Action are presented in Chapter IV. The elements of Deontic Logic are not treated until Chapters VIII and IX.

In my 1951 paper I took it for granted that the expressions which are formed of the deontic operators and symbols for acts can be combined by means of truth-connectives. This assumption would be warranted if the expressions in question could be safely regarded as the ‘formalized counterparts’ of sentences which express propositions. If, however, the expressions are also intended to be formalizations of norms, then it is not certain that the assumption is warranted. Propositions, by definition, are true or false. Norms, it is often maintained, have no truth-value.

The question whether norms are true or false challenges the question, what norms are. It is readily seen that the word ‘norm’ covers a very heterogeneous field of meaning, that there are many different things which are or can be called by that name. These things must first be classified, at least in some crude manner, before a discussion of the relation of norms to truth can be profitably conducted. This I have tried to do in Chapter I. One of the many types of norm which there are, I call prescriptions. After a more detailed analysis of the structure of norms, with the main emphasis on prescriptions, in Chapter V, the discussion of norms and truth is resumed in Chapter VI. No attempt is made to settle the problem for all norms. The view that prescriptions have no truth-value, however, is accepted.

The deontic sentences of ordinary language, of which the expressions of deontic logic may be regarded as ‘formalizations’, exhibit a characteristic ambiguity. Tokens of the same sentence are used, sometimes to enunciate a prescription (i.e., to enjoin, permit, or prohibit a certain action), sometimes again to express a proposition to the effect that there is a prescription enjoining or permitting or prohibiting a certain action. Such propositions are called norm-propositions. When expressions of deontic logic are combined by means of truth-connectives we interpret them as sentences which express norm-propositions.

The conception of deontic logic as a logic of norm-propositions challenges the question, what it means to say of prescriptions, or of norms generally, that they exist. Wherein does the ‘reality’ of a norm lie? This is the ontological problem of norms. Some aspects of it, relating chiefly to the existence of prescriptions, are discussed in Chapter VII. I find the problem extremely difficult, and do not feel at all satisfied with the details of my proposed solution to it. But I feel convinced that, if deontic logic is going to be anything more than an empty play with symbols, its principles will have to be justified on the basis of considerations pertaining to the ontological status of norms.

I still adhere to the opinion of my original paper that iteration of deontic operators to form complex symbols, such as OO or PO or O ~P, etc., does not yield meaningful results. Some kind of ‘iteration’, however, is certainly possible. For there can be prescriptions (and maybe norms of other types too) concerning the obligatory, permitted, or forbidden character of acts of giving (other) prescriptions. In a symbolic language, which contained expressions for such norms of higher order, deontic operators would occur inside the scope of other deontic operators. No attempt is here made to develop the adequate symbolism. But some problems concerning higher order norms (prescriptions) are discussed informally in the last chapter (X) of this book.

The building of a Deontic Logic has thus turned out to be a much more radical departure from existing logical theory than I at first realized. The more I have become aware of the complications connected with the subject, the more have I been compelled to narrow my claims to be able to treat it in a systematic and thorough way. What is here accomplished, if anything, covers only a small part of the ground which has to be cleared before Deontic Logic stands on a firm footing.

The main object of study in this book is prescriptions. Originally, I had planned to include in it also a fuller treatment of that which I call technical norms about means to ends, and the closely related topic of practical inference (necessity). But I have come to realize that this is an even more extensive and bewildering conceptual jungle than the topic of prescriptions. I therefore eventually decided not to attempt to penetrate it here. But I think that a theory which combines a logic of prescriptions with a logic of practical necessities is an urgent desideratum for the philosophy of norms and values.

I have lectured on norms and deontic logic both before and after my Gifford Lectures in 1959. I wish to thank my classes collectively for the stimulating opportunities which lecturing has given me to present ideas—often in an experimental and tentative form. In particular, I wish to thank two of my colleagues individually. These are Professor Jaakko Hintikka, whose criticism has effected profound revisions of some of my earlier views in the Logic of Action, and Mr. Tauno Nyberg, by whose advice and assistance I have greatly profited in preparing these lectures for publication.


From the book: