The Ascription of Actions o

Hart’s seminal paper, "The Ascription of Responsibility and Rights" 1 contains two insights, one about the logic of argument, the other about the structure of our concepts. He maintains that argument is, typically, not a matter of proving necessary and sufficient conditions but of establishing a prima facie case, considering and meeting objections and rebutting counter-arguments. Many of our concepts, not only in the law but in ordinary life, are shaped both by our practical needs and by the underlying logic governing their use. The right to property and, more importantly, the responsibility for actions is something that is owned by people, and in order to understand rights and responsibilities we need both to consider why people are allotted them and to elucidate the arguments for their ascription in particular cases. There has been some subsequent criticism of Hart’s paper, but these criticisms seem to miss the main issues, and although on some points Hart’s position needs to be modified or stated more carefully, his main contentions come through unscathed.

Geach condemns ascriptivism. He thinks that it is like prescriptivism, and impugns the truth value of statements about rights and responsibilities. Clearly, Hart would be wrong if he maintained that statements of the form "This is mine" or "He did it" could not be true or false in a fairly straightforward way. But the fact that the general purpose of such statements is to claim or ascribe a right or responsibility does not prevent our evaluating some particular claim or ascription as true or false. My statement that this car is mine would be false if in fact it was yours, and you had not sold, lent or hired it to me, and my saying that he did it may be a proper discharge of my duty to tell the court the truth. Hart does not deny this. He allows that a claim or accusation can be challenged by a denial of the facts upon which they are based (technically called a traverse or joinder of issue) 2 and that they can have straightforward descriptive uses, 3 but he does direct our attention away from these straightforward cases, partly in order to concentrate on the philosophically more interesting cases where the situation is not clear-cut but inherently "butty", partly because he is at pains to stress the performative aspect of some legal uses of ascriptive words, partly because he tends to assume that there is a sharp distinction between descriptive statements of fact which can be characterised as true or false, and interpretative utterances of law where the words ‘true’ and ‘false’ are inappropriate.

In so far as Hart assumes that there is a single sharp distinction between fact and interpretation, he is wrong; but such an assumption is unnecessary for his argument. The distinction between fact and interpretation is relative, not absolute, depending on the question at issue not some fixed philosophical schema. Often that a particular field is a particular person’s property will be a simple description of one of the facts of the case; and occasionally whether a sample is a piece of earth will need chemical analysis and be a matter of dispute between biologists. But this does not weaken Hart’s analysis; on the contrary, it shows that it is not limited to legal and quasi-legal concepts. Again and again our philosophical perplexities are due to our not realising that our concepts are on different levels, so that necessary and sufficient conditions for the application of those of one level cannot be defined in terms of concepts from some other level. We do not have to confine the words ‘true’ and ‘false’ to statements involving only concepts of some specially basic level. Although sometimes we are chary of describing judgements as true or false, either because they are performative or because some other evaluation seems more apt, Hart’s ascriptivism does not in general preclude our characterising statements about property or actions as true or false.

Pitcher offers a number of counterexamples to Hart’s thesis. Some ascriptions of action are not, as Hart maintains, defeasible, but can be conclusively justified by reference to bodily movements, with no possibility of defeating the ascription by reference to further facts. This indeed is sometimes true, just as sometimes in the law we lay down strict liability, and hold a man responsible for certain states of affairs in spite of all extenuating circumstances. The verb ‘to bump’ is more firmly tied to one’s bodily behaviour than the verb ‘to hit’. If my hand or arm comes into violent contact with your body, it establishes only a prima facie case that I hit you, and I can defeat the accusation by showing that I was trying to give you a helping hand, or that I was pointing out an interesting bird, or that I did not know you were there. But none of these circumstances would defeat the statement that I bumped (into) you. Bumping discounts the excuse of accident, whereas hitting involves some element of intention. Our action words are not, as Hart seems to suggest, all of one sort, but, rather, range from those that do little more than characterise bodily behaviour to those which are more concerned with the reasons for acting than with overt movements of anyone’s body. But this does not diminish the value of his paradigm. We need to refine his thesis to take account of Pitcher’s examples, but it will be only an elaboration of his original insights.

Why do we need to establish the ownership of actions? Traditional accounts of responsibility are too blame-centred. People are required to own up to their misdeeds in order that they can be blamed for them. But this is a distortion. Naturally, we are most energetic in disclaiming responsibility when we are in danger of disapprobation, and therefore lawyers earn their fees in fighting ascriptions of ill-doing, but in ordinary life we are concerned not so much to blame as to understand. I want to know who is responsible, answerable, in order to have him answer my question "Why did you do it?" Only he can give me a first-hand account of the rationale of action, and only if I know that can I really understand what was going on. If the reasons were disreputable, the agent may be reluctant to acknowledge them, and may seek to evade my question "Why did you do it?" by claiming that it is not a question he can be expected to answer, because he is not responsible for what happened. There are many circumstances in which responsibility can properly be disclaimed, and Aristotle and Hart have shown how heterogeneous these can be: but sometimes we are tempted to go further, and represent ourselves as not being responsible for our actions simply because we are not worthy of blame, and instead of avowing our real reasons which would in fact exculpate us, make out that our actions were in some sense involuntary, and therefore not really ours at all. 4 We should avoid confusion by distinguishing the two questions, one concerned with who should be asked to say what his reasons were, and the other concerned with whether in view of those reasons and what actually happened, he is worthy of censure or commendation, punishment or reward.

Reasons issue in results, but the relation between them is complex. Typically, although not always, an action is characterised by reference to the reasons for which it was undertaken, and to say what one is doing is partly also to say why one is doing it. Often, of course, we have difficulty in articulating our reasons adequately in words, and usually where other people are concerned we have to read their reasons in their behaviour. Behaviour, because it is overt, is readily available as a public criterion of a man’s actions, and it is natural in a public language to be very largely guided by publicly observable behaviour in our ascription of actions to others. If a man’s hand collides with my cheek, I construe it as a blow, and am justified, in the absence of further circumstances or explanations, in saying I have been not only hit but insulted. But explanations alter the complexion of the case. If my apparent assailant explains that a horsefly had just settled on my cheek, we should hesitate in describing his action as hitting me, although my cheek would smart just the same. If it was part of a play, or had been stage-managed to divert attention, then it would be true to say that he had hit me, but not that he had insulted me. The same pattern of bodily behaviour might be correctly described in these three, as well as many other, ways. In the absence of further circumstances or explanations, an onlooker would be justified in jumping to the most adverse conclusion, although the possibility remains open of the action’s being described in much milder terms; that is to say, the overt bodily behaviour constitutes adequate, although not conclusive, grounds for the action’s being described as a man’s insulting another. The concept of action is, typically, open-textured. Actions can be confidently described by someone who has witnessed them, even though on occasion he may, in the light of further circumstances or the agent’s own avowals, have to withdraw his original description and characterise them in quite different terms. But too much openness would be intolerable. The concept of action would lose all content if it were so open that a man could always disclaim responsibility by saying ‘I did not mean to’, and since it is always easy to say ‘I did not mean to’, we sometimes, especially with actions whose consequences can be foreseen to be grave, import a definitional stop, and make certain pieces of bodily behaviour conclusive for the ascription of certain actions. If I take a gun, load it and pull the trigger, I cannot say I did not mean it to go off. I have fired the gun, and nothing can defeat that description of my action. Even when no serious consequences are involved, there is a natural tendency in a public language to make the publicly observable criteria conclusive, and we are inclined to say "He hit her but he did not mean to" rather than "He did not hit her, he only bumped into her." These pressures combine to produce the examples Pitcher cites. We should note them and accommodate them, but not take them as refuting Hart’s basic account.

Hart elucidates rights as well as responsibilities, and it is one of the merits of his account that it reveals the structure of argument, not only in the law, but in morals, history and the humanities generally. Philosophers have too long construed argument more geometrico, as a monologue in which one goes from premisses to conclusions in virtue of certain rules of inference which can be seen to apply and which clearly license the transition. Mathematical argument can be reconstructed to fit this pattern, though even then something is lost by representing argument as always a monologue; 5 but most sorts of argument are essentially dialogues, arguments between two persons, in which various considerations are adduced to support each side, and once an adequate case has been made out, it will be successful in default of a convincing counter-consideration on the other side. We do not look for arguments that are conclusive, but only for arguments that are adequate. We do not seek to show that it is impossible to gainsay our view of the matter, but only that on balance the weight of argument favours it. We deal not with definite clear-cut contentions which prove their point by knock-down arguments beyond all possibility of cavil or rejoinder, but attempt to articulate all sorts of considerations, often with opposite tendencies, and decide after due deliberation where the answer lies, recognising that our judgements are fallible and our tentative conclusions could be shown wrong by some further factor.

Few people would deny Hart’s negative contention that the logic of the logic books does not adequately represent the pattern of actual argument whether in the law courts, or in practical life, or in many academic disciplines. The actual schema Hart offers instead is only very sketchy. This in my view is a virtue, and betokens a properly non-procrustean attitude towards the actual exercise of human reason, but some philosophers may wonder whether Hart’s positive contentions are actually borne out by the facts. As regards the adversative nature of argument, I shall only cite the very large number of adversatives currently employed in academic English – but, however, nevertheless, notwithstanding, on the other hand, whereas, while, even so, although, yet still, in spite of, – compared with the smaller family of therefore, so, if and because. The open-texturedness of adversative argument is a corollary of its adversative character, but more difficult to establish by independent examination. We can infer that there must be something in the argument from silence in virtue of the caution urged against relying too much on it. Sir David Ross, meditating on the nature of moral argument, 6 was led to characterize the typical moral argument as "prima facie", and cautious academics often have occasion to insert a ceteris paribus, or ‘other things being equal’ clause, or qualify their conclusions as holding only in the absence of countervailing considerations. But once we accept the adversative character of actual argument, we are allowing that there is in general a possibility of a further ‘but’, and where this possibility is not actualised can argue further from the failure of our adversary to rebut our contention, even though he can be presumed to be anxious to rebut it if he can, that it is indeed un-rebuttable.

The structure of actual argument has not yet been adequately elucidated, although Toulmin and Pole have made a useful beginning. 7 Although we cannot hope to formalise it to anything like the extent that mathematical logic has been formalised, we might find it helpful to compare the two approaches. The mathematical logician thinks in terms of a function with many, but only finitely many, "arguments" (in the mathematical sense – here, in effect, the premisses in the logician’s sense) whose value is the conclusion; or, alternatively, in terms of a relation between a finite set of premisses and a single conclusion. Except for sets of premisses that are together inconsistent, no subset of the premisses can yield a conclusion inconsistent with that yielded by the whole set. It is essential that we always know what the premisses are and can usually tell whether or not the conclusion follows. It is essential to Hart’s account that we cannot know in advance what all the factors are. We know only an initial segment of the relevant considerations, and the conclusions suggested by one initial segment may be at variance with what had been suggested by a subsegment of it. To take a very simplified model, we may think of the infinite sequence of possibly relevant factors as analogous to an infinite decimal specifying a real number. For Hart, the question whether all the circumstances of a concrete situation support a certain conclusion is like asking whether a particular real number has a certain characteristic, whereas for the mathematical logician we have to ask the question not of particular real numbers but of particular intervals of real numbers, and after some finite number of decimal places we shall know that every real number with that decimal expansion (and therefore every real number within the corresponding interval), has, or perhaps does not have, the given characteristic. Against the relative lack of structure imposed by Hart’s account, we have certain other requirements we ordinarily impose on actual arguments, some requirement of relevance, sometimes a requirement of universalisability, and a ban on special pleading. 8 These bear some similarity to the requirement of continuity in functions of real numbers. We must treat like cases alike, and although we cannot lay down in advance what constitutes cases being alike (specify in advance an interval of real numbers all being for our purposes exactly the same), but must always leave open the possibility that circumstances alter cases (allow that within any given interval two real numbers may have, with respect to some given function, significantly different values), nevertheless we can require that for any given degree of similarity between the values of a given function there must be some interval round a real number within which every number has a value that is itself within that degree of similarity (we require that if we are to discriminate between cases there must be some distinguishing feature to justify our differentiating between them).

Rebarbitative though this model is to the unmathematical reader, it is over-simplified. It attaches too much significance to the order of premisses or factors. Particularly for the mathematical logician, no importance should be given to the order of his premisses. We do not need to deal with initial segments, but simply subsets, and if we are using initial segments as a model, we should need to have it that any two intervals were equivalent if their initial segments contained exactly the same numerals irrespective of order. It is also inadequate in having only a small number of numerals. There are many more than ten possibly relevant factors; indeed, indefinitely many. Although the mathematical logician operates with only a finite number of premisses in any actual argument, there are a denumerably infinite number of propositions which could be taken as premisses. The mathematical logician is operating with finite subsets of a denumerably infinite set, with no account of order, but subject to the rule that if some finite set of premisses yields a particular conclusion, then no subset can yield a conclusion inconsistent with it, whereas to obtain a Hartian model we have to think in terms of sets not necessarily finite and not subject to a subset consistency rule. But although we cannot avail ourselves of the simple model of the real numbers, we may still be able to devise a model that captures the crucial features we wish to exemplify. Topology started with the analysis of the real number continuum, and has been gradually weaned from its numerate parentage; and, granted some distinction between general features and particularised descriptions analogous to the distinction between open and closed sets, I can give at least a partial analogue of our Principle of Universalisability and our ban on special pleading. Much more needs to be done. We cannot hope to give a complete formalisation of adversative argument, but I hope that someone sufficiently accomplished in manipulating the Boolean algebra of sets will be able to represent our main intimations of what distinguishes a valuable progression of arguments from a barren interchange of debating points.

Mathematical models by themselves carry little conviction, and whether or not an illuminating model of adversative argument can be devised, many philosophers will be guided by quite other considerations in deciding whether to accept or reject Hart’s account. Even if he accurately portrays the course of actual argument, philosophers may still feel some metaphysical disquiet at arguments being represented as activities rather than timeless logical relations, and at our canonizing the fallible and provisional judgements of non-omniscient men rather than the incontrovertible and conclusive inferences of traditional logic. But it is better suited to our human condition. We are characteristically having to decide in situations of imperfect information. We are rational agents. As agents we are always having to make our minds up about what to do, and shall be lost if we hesitate too long. I cannot dither, but must decide either to stand and fight or to turn and flee while yet there is time. So too in the law courts tertium non datus, the judge must find either for the plaintiffs or for the defendants. It is a rare luxury, mostly confined to academic life, to be able to suspend judgement. Usually we have to reach a decision on the evidence available to us at the time, even though the evidence is inadequate and our judgement faulty. All we can do is to act not merely as agents but as rational agents, and make the best decision we can. Although in most cases the evidence does not indicate one course of action to the exclusion of every other possibility, it is inherent in the idea of a rational agent that our actions should be a response to the situation in which we find ourselves, and that some actions are more appropriate in a given situation than others. Hence it is that the situation, as we see it, can constitute a prima facie argument for some decision or other, but not a conclusive one. Hence also the fact that further circumstances can alter the complexion of the case. If our information is relevantly imperfect, it follows that it may become less imperfect, and that this may be relevant, i.e. that our decision may be different in the light of it. And so both in the practical decision-making of ordinary life and therefore also in the adjudications of the law courts, it is inherently possible that, although all the circumstances on which a decision could be based are present, yet in the particular case that decision should not be taken because other circumstances are present which casts a different light on the case. Because we are agents, we have to take decisions as occasion arises. We can do no other. Because we are rational agents, we can be right, but, by the same tokens, can also be wrong. Because we can be wrong, our arguments are open to correction and amendment, and so should be seen in the context of a dialogue, not a monologue, and our concepts are similarly defeasible and open-textured.


I wrote this paper in the 1970s as possible contribution to the Festscrift for H.L.A.Hart, Law, Morality and Society edited by P.M.S.Hacker and J.Raz, Oxford, 1977. It did not suit, as Gordon Baker was writing on the same topic, and I put the paper away. I did send it to Hart, but it did not lead to his agreeing to republish his article. There is some discussion of this in N.Lacey, A Life of H.L.A.Hart, Oxford, 2004, pp.146, 376-377; and in consequence I was asked by Dr S. van Hoorn, of Nijmegen University, to put it on the web.

1.Proceedings of the Aristotelian Society, 1948, pp.171-194; reprinted in A.G.N.Flew, ed., Logic and Language, Oxford, 1951, ch.VIII, pp.145-166; page references are to the reprint.
2. p.147
3. pp.156-159
4. See, especially, Aristotle, Nicomachean Ethics, III, 1, 4-6, 1110a4-18.
5. See J.R. Lucas “Not ‘Therefore’ but ‘But’”, Philosophical Quarterly, 1966, pp.289-307.
6. W.D. Ross, Right of the Good Oxford, 1930, pp.19ff. and Foundation of EthicsOxford, 1939, pp.84f.
7. S.E. Toulmin, The Uses of Argument, Cambridge, 1958; and David Pole, Conditions of Rational Inquiry, London, 1961.
8. See J.R. Lucas ``The Lesbian Rule'', Philosophy, 30, 1955, pp.195-213.

Click here to return to home page

Click here to return to bibliography