The Suntory and Toyota International Centres for Economics and Related Disciplines
The Logic of Surprise
Author(s): G. L. S. Shackle Source: Economica, New Series, Vol. 20, No. 78 (May, 1953), pp. 112-117 Published by: Blackwell Publishing on behalf of The London School of Economics and Political Science and The Suntory and Toyota International Centres for Economics and Related Disciplines Stable URL: http://www.jstor.org/stable/2550836 Accessed: 26/04/2010 18:23 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Blackwell Publishing, The London School of Economics and Political Science, The Suntory and Toyota International Centres for Economics and Related Disciplines are collaborating with JSTOR to digitize, preserve and extend access to Economica. http://www.jstor.org [MAY The Logic of Surprise By G. L. S. SHACKLE A SURPRISE IS EXPECTED ". These words on a newspaper placard carry at first sight the suggestion of a contradiction in terms. Surprise is what we feel when an expectation has gone wrong, has proved in the event to have been ill-founded and false. How then can one expect to be surprised ? One possible explanation is trivial: the expectation was in the mind of person A and concerned a feeling to be experienced by person B. But let us ask whether in any sense a man can expect that he himself will on some specific occasion experience surprise. On broad grounds of common experience and common sense we may well be inclined to start by saying: Well, we all know that there is such a thing as a feeling of surprise, we have all experienced it at one time and another, and it would be absurd to say to oneself at any stage of life: " I shall never again be surprised ". 'So at least in a vague and general sense we do assume, and accept the fact, that life has surprises in store for us, though whether this is more than the inescapable result of cold reflection, whether it really represents a lively and insistent conviction, is perhaps more doubtful. But plainly there is here a paradox: on the one hand there is the fact that each one of us feels sure that he will in the future experience surprise, on the other hand there is the fact that surprise is what we feel when that has occurred which we did not expect. It may be asked: Is the indefiniteness of the future occasion, its being quite unspecified as to date or circumstances, an essential condition for the possibility of entertaining a belief that we shall feel surprise ? No: it is, I think, possible to expect to be surprised by what will happen on an occasion which is in some sense particularised. The paradox is a close-knit and definite one. To resolve it we need first some definitions. There are two things that I think we can reasonably mean when we say that a person expects such and such a state of affairs or such and such an event. The state of affairs being specified, and its location in future history being stated by naming a calendar date or by describing an organic sequence of situations of which the particular state of affairs is the end-product, or by naming some, so to speak, pigeon- hole in future history such as ' the next General Election ' or ' the next debate in the Commons on economic affairs', we can mean, when we say that a given person ' expects ' this state of affairs to arise at the named ' cue ', either that he attaches zero potential surprise to this as well as to many other mutually exclusive hypotheses, or else that he attaches some degree greater than zero of potential surprise 112 1953] THE LOGIC OF SURPRISE 113 to all hypotheses other than this one. Even with the first, and weaker, of these two meanings our paradox appears. In conversation we use such words as ' surprising' and ' unexpected' with little heed to fine shades of meaning or to precision and con- sistency. But for my present theme I must recall a distinction which I suggested some years ago1 and which is not, I should maintain, a fine or subtle difference of meaning but a plain and essential one. By a counter-expected event I mean an event which has figured as a member of the set of hypotheses provisionally entertained by some individual about the outcome of some course of action or some ' experiment', or about the answer to some question, and which has been carefully examined by him in that context and as a consequence has been assigned a high degree of potential surprise. A counter-expected hypothesis, that is to say, is in rough language a hypothesis that has been looked at and rejected. In contrast with this I define an unexpected event as one which has never been formulated in the individual's imagination, which has never entered his mind or been in any way envisaged. But here we must be careful. Evidently the question whether an event has been thought of or not may depend on the degree of exactness and detail in which it is specified. If I lose my latch-key, the possibility that I shall find it 'at home' will no doubt be in my mind; but the idea that it could have found its way into my three-year-old daughter's money-box may have escaped me in that explicit form. Yet this also answers to the description of finding the key ' at home'. These two categories, counter-expected and unexpected situations or events, are exhaustive in the sense that under one or other of these heads can be included every kind of event or state of affairs which does not conform fully and precisely to an imagined situation or event pre-existing in the individual's mind and carrying zero potential surprise. Now to say that someone expects to be surprised by a counter-expected event involves a logical contradiction. Some precisely specified future event is imagined, and some degree, zero or greater than zero, of poten- tial surprise is attached to the hypothesis of its occurrence. To expect this event means, at any rate, to assign to it zero potential surprise; but in that case this zero degree of surprise is what is expected to be experienced if this event occurs. Thus it follows that our paradox can only be resolved by supposing that one can expect to be surprised by an unexpected event. In what circumstances will this be possible ? This latter question is really two questions: (a) What circumstances expose a man to the occurrence of an unexpected event ? (b) In what circumstances will he be aware of this exposure ? Here we are not thinking, of course, of events in general, but of the outcome of specified antecedents, the result of an ' experiment' defined 1 "The Expectational Dynamics of the Individual ", Economica, May, 1943. See also Expectation in Economics, Cambridge University Press, 1949, pp. 73, 74, 76. 114 ECONOMICA [MAY in some respects as to its character and particular occasion. All of us, needless to say, are at all times exposed to unexpected occurrences in general. Question (a) asks in what circumstances, when a man has been looking at some particular pigeon-hole of future history, what actually happens can have altogether escaped his survey of possibilities, so that the degree of potential surprise he assigned to it was neither zero nor greater than zero, but was non-existent, a sheer blank. If the question to which the event will provide the answer is a very simple one, so that the answer will be, for example, simply Yes or No, or a single number, then it is plainly almost impossible that the individual should fail to survey, at least in principle and by implication, all out- comes which are logically conceivable, so that no loophole will be left for anything to occur which he has not thought of and to which he has not assigned some degree of potential surprise. When the question is a Yes or No question or a ' one-dimensional' question, there can scarcely be any possibility of an occurrence which is unexpected in my special sense. It is when the answer must consist in a catalogue of a great many particulars and details that the very existence of some of the individual characteristics, or the possibility of certain combina- tions of them, may be overlooked by the individual when he considers the matter in advance. Complexity of the event is therefore, I think, a circumstance which must be present if the individual is to be in fact exposed to the occurrence of an unexpected event. Thus, for example, the size of a dividend could not be unexpected in my sense, though it could of course be counter-expected. But a Budget, or some of its proposals, could be unexpected, because the variety of possible com- binations of measures is so great. We come, then, to question (b). By an exhaustive set of rival hypotheses about some question I mean a set of headings under one or other of which the individual feels certain that he will be able to place the true answer to the question when it shall become known. Some at least of these headings will ordinarily be stated with some precision and fulness of detail, but the individual may not feel certain that the true answer will be found amongst these accurately stated hypotheses. If so, in order to make the set of hypotheses exhaustive he will have to include a residual hypothesis, which will be simply a recognition of the non-exhaustiveness of his list of precisely stated hypotheses. Since, if the question is meaningful, it must necessarily have some true answer, the individual will acknowledge that one at least of his exhaustive set of rival hypo- theses must be assigned zero potential surprise. But there is nothing in general which compels him to attach zero potential surprise to any of his fully stated hypotheses: every one of these may carry potential surprise greater than zero, and the zero degree be attached only to the residual hypothesis. If so, there is no logical contradiction in his believing it possible that when the true answer shall become known, its character in detail will surprise him. More precisely we can say that no logical contradiction is involved in the individual's assigning 1953] THE LOGIC OF SURPRISE 115 zero potential surprise simultaneously to each of the ideas (1) that the true answer when it shall become known will fall under his residual heading, and (2) that its character in detail will cause him surprise. By way of illustration, suppose that a 19th century physicist had been shown an electronic digital computer performing in a matter of minutes calculations that no human computer could execute in a lifetime. Unable to conceive how the miracle was wrought, yet convinced that the apparently impossible was in fact being done, he might well have attached potential surprise greater than zero to any conjectured, but plainly unsatisfactory, explanations that had occurred to him, and so have been compelled to reserve zero potential surprise for a residual hypothesis. And he might well have admitted to himself that the true explanation when it should be revealed was almost bound to surprise him. It is evident that the existence, amongst his exhaustive set of rival hypotheses, of a residual hypothesis is a necessary condition of his being able to attach zero potential surprise to the idea that the eventual true answer will surprise him. For if all the hypotheses which together compose the exhaustive set were each fully specified in detail, he would, by the meaning of the words, not expect to be surprised by any one of those which he expected to occur. But is the existence, in his ex- haustive set, of a residual hypothesis a sufficient condition for him to ' expect to be surprised ' ? I think the very fact that he has been baffled in his attempt to formulate in detail an exhaustive set of rival hypotheses indicates that he will necessarily be surprised by the truth which resolves this bafflement: anything not capable of surprising him would have occurred to him. If we say that the existence of a residual hypothesis is a sufficient (as well as a necessary) condition for the individual to be able without logical contradiction to ' expect to be surprised ' by the true answer, we are giving to the word ' expect ' the stronger of the two meanings mentioned above. For we are saying that the individual will attach potential surprise greater than zero to the idea that the true answer, when it shall become known, will not surprise him; we are making the word ' expect' mean that the in- dividual has some positive degree of confidence in the idea that he will be surprised. The argument to this point may be summarised thus: an actual experience will cause surprise when an imperfect image of it has been formed in advance; and the imperfectness can consist either in some wrong characteristics or details being specified in place of right ones, or in blanks having been left in the picture, that is, it can consist in characteristics or details which in fact belong to the experience having been omitted altogether; or again we can say, in the experience having ' dimensions 'beyond the list of those which the individual could specify in advance. If the existence of these blanks, and the individual's inability to fill them in, has been recognised by him in advance, and if 116 ECONOMICA [MAY he also recognises that their existence exposes him at least to the possi- bility of being surprised when the actual event fills them in, he can be said to ' expect to be surprised '. For he will then attach zero potential surprise to the hypothesis that the filling of the blanks will, by its manner or matter, not by its occurrence, surprise him; and to attach zero potential surprise to a hypothetical future event is one thing that we can mean by ' to expect it'. Our argument regarding the possibility of ' expecting to be surprised' has had as one of its central ideas the concept of residual hypothesis. The part that this idea can play in the analysis of real situations may be illustrated in conclusion of this note. We have seen that the in- dividual will only be driven back on a residual hypothesis when the question: What will be the outcome of this course of action ? presents itself to him in a rather complex form, calling for an intricate schedule of information rather than a single number or a simple Yes or No. But it may be that this complicated kind of answer is going to be treated as a means to a simpler kind, which will sum up in some values of a single variable the happy or unhappy possibilities implied by the complex answer. The values of this variable corresponding to different hypothetical answers of the complex kind, each associated by the individual with some degree of potential surprise, will constitute a set from which he can select two focus outcomes which, when standardised, he can use as co-ordinates to plot a point, representing the course of action in question, on his gambler indifference map. Now it seems evident that the only way in which a residual hypothesis concerning the complex type of question can be ' summarised ' by means of a single variable is by two very widely separated values' of that variable, one of them representing a highly desirable and the other a highly undesir- able outcome. If two such values select themselves as the focus values of a course of action concerning whose detailed outcome a residual hypothesis is entertained, the point representing this course will lie far from the origin of the gambler indifference map and, if losses are measured on its horizontal axis, far towards the right where focus losses approach the maximum possible size represented by the in- dividual's whole fortune. Now there is reason to suppose that the slopes of the gambler indifference curves on such a map become steeper towards the right. Thus a point whose location is determined by a residual hypothesis is very likely to lie below the origin indifference curve, that is, the gambler indifference curve on which all points are esteemed by the individual as neither more nor less desirable than an assurance of neither gain nor loss, a situation which, in the context of gains or losses measured in money, he can attain by simply keeping his whole fortune in the form of a bank-balance. However, if only some of his possible courses of action have a residual hypothesis, does the conclusion we have just reached have any practical bearing ? 'Their separation can of course only be meaningfully described as ' wide ' in relation to a particular context. 1953] THE LOGIC OF SURPRISE 117 The state of mind in which the individual can only render exhaustive his set of hypotheses concerning some one or more questions or courses of action, by including in each such set a residual hypothesis, can arise in two ways. The feeling of unplumbed ignorance or lost bearings can be concerned with the situation of which some particular actions of the individual would be the 'proximate cause'. Or it can be concerned with the framework of events which will happen ' in any case ', the things which are altogether outside his control, such as the weather or the political situation. In the latter case it may well happen that not one of the more adventurous of the courses of action open to him escapes the need of a residual hypothesis to render exhaustive his set of rival hypotheses concerning its outcome. If he attaches low or zero potential surprise to these residual hypotheses, all the points represent- ing active and ' positive' policies may lie below the origin indifference curve, and only those representing passive policies, typically that of keeping his fortune in cash, may lie on or above it. Thus we should expect that when the news seems to provide internally inconsistent or confficting evidence about the policies or intentions of powerful persons, or concerning the degree of knowledge they possess, or is for any reason especially difficult to interpret, the consequence would be to inhibit some kinds of business activity: not because the news was regarded as bad from the point of those engaged in this activity (often they are divided into two camps with opposing momentary interests, like the Stock Exchange bulls and bears) but because it is unintelligible. And this we actually observe to happen. University of Liverpool.