Back tothe SyllabusGeorge C. Homans, Elementary Forms of Social Behavior, (2ndEd.), New York: Harcourt Brace Jovanovich, 1974.
Our first proposition relates a man's (or woman's) action to itssuccess in getting a favorable result. In classical psychology it iscalled "the law of effect." Because we believe another name will makeits meaning more obvious, we shall call it the successproposition.
We may state it as follows:
I. For all actions taken by persons, the more often aparticular action of a person is rewarded, the more likely the personis to perform that action.
The proposition in itself says nothing about the reasons why theperson performed the action in the first place. In the case of anexperimental animal like a pigeon, its repertory of innate behaviorseems to include a tendency to explore or investigate its environmentby pecking at the objects within it. The psychologist may have soarranged its cage that the motion of a metal key will release a grainof corn to the pigeon. If, in the course of exploring its cage, thepigeon happens to peck at the key and thus gets the corn to eat, theprobability that the pigeon will peck the target again will increase.Not until then will the psychologist be able to use the pigeon'stendency to repeat its action for the purposes of furtherexperimentation. The same sort of behavior is characteristic of men.What the success proposition says is that, whatever be the reason whya person performs an action, once he has in fact performed it and theaction has proved successful the result has for the person what weshall later call positive value-- then the person is apt to repeatthe action.
The result of an action is what follows it. The successproposition holds good even if success was not, in the eyes of someinformed observer caused by the action but was rather a matter ofchance. Much of the magic men have performed has been maintained byfortuitous success, especially when success is much desired andalternative means of producing it are not known. After all, rainusually does follow the magic of rainmaking-- sooner or later.
The proposition may sound as if it said that an action were causedby its result, which is absurd to those of us who do not believe inteleology. But it does not say that. What we observe is a sequence ofat least three events: (1) a person's action, which is followed by(2) a rewarding result, and then by (3) a repetition of the originalaction or, as we shall see, by an action in some respects similar tothe original. It is the combination of events (I) and (2) that causesevent (3), and since the former two precede the latter in time, weare saved from teleology. It is natural to call the original sequenceof three events a learning process, and therefore the generalpropositions we shall use are often called the propositions of"learning theory." We believe this to be a mistake, since thepropositions continue to hold good long after the behavior has inevery ordinary sense of the word been learned.
The fact that a person's action has been rewarded on one occasionmakes it more probable that he will repeat it on the next occasion.If there are many such occasions, the probability that he willperform the action will vary directly with the frequency with whichit has been rewarded, and we have deliberately cast the propositionso that it takes this form. Remember that we are particularlyconcerned in this book with the process by which social behaviorgives rise to relatively enduring social structures. Without repeatedsocial actions there are no enduring social structures.
The proposition implies that an increasing frequency of rewardleads to an increasing frequency of action, but it is obvious thatsuch an increase cannot go on indefinitely. It has built-in limits,as we shall see later when we consider satiation. The propositionalso implies that the less often an action is rewarded, the lessoften it is apt to be repeated. At the extreme, if an action oncerewarded is never rewarded thereafter, a person tends in time neverto perform it at all. In the technical language of behavioralpsychology, it eventually becomes extinguished. But the timerequired for extinction may be very long indeed, and a singleoccasion on which the action is rewarded may be enough toreinstate it at full strength.
Let us now consider some qualifications of the successproposition. The shorter the interval of time between the action andthe reward, the more likely the person is to repeat it-- the morelikely, to use the language of everyday life, he is to "see" theconnection between his action and its reward. If we wish a person tolearn, we shall do well to reward his correct responses promptly.This is the principle on which "teaching machines" are based. Thereason why we do not use ordinary language but a proposition whichmerely sums up the facts is that everyday language is apt to embodyassumptions about human behavior that are not always justified. Thusprompt reward is apt to make action more probable even if the persondoes not "see" the connection between his action and its reward inany conscious sense. The greater, moreover, is the value of thereward, the more likely is the person to make the connection, but weshall have much more to say about value later.
The frequency with which the person performs the action dependsalso on the pattern in which the reward comes. (On this matter seeespecially Ferster and Skinner, 1957.) For a given total number ofrewards within a given period of time, it looks as if a man, like anexperimental animal such as a pigeon, will repeat an action lessoften if it is rewarded regularly--for instance, if it is rewardedevery time it is performed-- than he will if it is rewarded atirregular intervals of time or at irregular ratios between the numberof times he performs the action and the number of times it isrewarded. Furthermore an action once regularly rewarded will, whenthe reward ceases, become extinguished sooner than one rewardedirregularly. One reason why people are willing to work so hard atgambling, fishing, or hunting, even when they have little success, isthat such actions are characteristically rewarded irregularly. Indeedthe tendency to repeat an action more often if its reward comesirregularly may have arisen in animals, including the ancestors ofmen, because of its survival value. If one depends for one's food onactivities such as fishing and hunting, one had better not give uptoo easily if one is unsuccessful, but persist. The tendency impliesthat animals will do just that.
Though we take note of these relationships subsidiary to thesuccess proposition, we shall have little more to say about them.They do not render invalid the success proposition itself. Even inits crude form, the latter holds good over a wide range of behavior.In gross and in a first approximation it will serve us well inexplanation.
We turn now to the second of our general propositions-- butremember that they are "our" propositions only in the sense that weuse them and not that we discovered them. This proposition concernsthe effect on action of the circumstances attending it. Since in manyaccounts of operant or voluntary behavior these attendantcircumstances are called stimuli, we call this the stimulusproposition.
We may state it as follows:
II. If in the past the occurrence of a particular stimulus, orset of stimuli has been the occasion on which a person's action hasbeen rewarded, then the more similar the present stimuli are to thepast ones, the more likely the person is to perform the action, orsome similar action, now.
In formulating their theories some psychologists include thereward of the action itself among the stimuli, referring to it as areinforcing stimulus. We believe it is confusing to do so. It is truethat the sight of some object that we have coveted and obtainedearlier is a stimulus to our efforts to obtain it again; but it isthe sight of the object, not the success in obtaining it, that is thestimulus. If we confuse the two, proposition (II) would seem to saythe same thing as proposition (I), whereas they really say somethingdifferent.
Proposition (II) says that the reappearance of the circumstancesattending successful action make more probable the repetition of thgaction. Thus a fisherman who has cast his line into a dark pool andhas caught a fish becomes more apt to fish in dark pools again. Theconnection between the stimuli and the action is subject to bothgeneralization and discrimination. If our fisherman hasbeen successful in a dark pool, he may come to fish more often in anypool that is to some degree shady. Indeed his action itself maygeneralize. If he has been successful at one kind of fishing, he maybecome prepared to try other kinds and even other related sports,such as hunting. On the other hand, he may learn to fish only undervery specific conditions of water, light, and shade, provided he hasbeen successful under these but not under other conditions. In thiscase, the stimuli that govern his behavior have become highlydiscriminated. Should the conditions under which success is alonepossible become complicated, they may not establish themselves at allas stimuli for his action. He is, as we say, unable to recognizethem. As in the case of reward, the temporal relationship betweenstimulus and action makes a difference: if the crucial stimulusprecedes the action by too long a time, the actor may not make theconnection. The greater the value of the reward, the more sensitiveto stimuli the person may become--so much so that if the value to himof a potential reward is very high, he may become oversensitive and,until corrected by failure, respond to irrelevant stimuli. Finally,alertness to stimuli or attentiveness to stimuli is itself an actionwhich, like any other kind of action, a person may perform more oftenif it has brought him reward. All of these relationships should belooked on as subsidiary to the main stimulus proposition.
In social behavior persons and their attributes become crucialstimuli. Did this person, rather than another, reward a man'sactions? If he did, his identity was one of the circumstancesattending successful action, and his presence on some new occasion isa stimulus making it more likely that the man will once more directsimilar action toward him. Does this person display the cold blueeyes that a man's father did when the father punished him long ago?Then the grown man may show some slight tendency to avoid such aperson. In human social behavior, what complicates the stimuli evenmore is the fact that they are largely verbal. The use of languagesets the behavior of men further apart from that of animals than doesanything else. The same general propositions apply to the behavior ofboth, but within these propositions the complexity of the stimuliavailable to men in their interaction with each other make possible ahigher order of complexity in their behavior.
The crucial variable in the stimulus proposition is obviously thedegree of similarity between present stimuli and those under which anaction was rewarded in the past. Yet similarity may not vary along asingle dimension but along many, and indeed it may depend on acomplicated pattern of measures. The ways in which personsdiscriminate among, or generalize across, combinations of stimuli isthe subject of the field of psychology called perception orcognition. So various and so many are the findings in thisfield that in this book we shall only state the stimulus proposition,though we shall feel free to use more specific findings ad hocto explain particular cases. The real intellectual danger is not thatthe findings are complex but that some social scientists shouldbelieve perception and cognition to be essentially different fromother behavior and thus require a different type of explanation. Theyare not essentially different. The ways in which men perceive andthink are just as much determined by the results they achieve as areother kinds of behavior.
Yet we do not wish to appear so rigidly behaviorist as to denyreality to some processes in which perception plays a large part andwhich are of great importance in social behavior. Our view of theseprocesses has been greatly influenced by the work of Bandura (1969).
Men, like many animals, often imitate the behavior of others oftheir kind. Imitation of others naturally requires some degree ofobservation of their behavior. We believe that a tendency to imitateothers is genetically inherited and not initially learned throughoperant conditioning. Yet, whatever its origin, a man will notpersist in performing an action he has imitated unless that actioneventually brings him reward. If it is successful in bringing himreward, he will not only be apt to repeat it but also to adoptimitation as a generalized form of behavior. Then his practicalsuccess will support the genetic tendency; and the persons he hasimitated will become stimuli in whose presence he will be especiallylikely to carry out imitative actions again.
Evidence is also accumulating that men can learn to act in acertain way even when, at first, the reward they get from the act isonly vicarious. Suppose that a child sees another child put a boxagainst a wall and use it to climb successfully out of his yard. Atthe moment, the first child has no occasion for climbing a wallhimself, but if he does have such occasion later, the evidencesuggests that he is much more likely to look for a box than he wouldhave been if he had not observed the other child, even though he hasas yet received no reward himself. Naturally he will not go onrepeating the action unless sooner or later he is personally rewardedfor performing it, but the initial stimulus to the action is theobserved success of the other child, not his own. This kind oflearning has been called model learning. The success of anyone action originally modeled on the action of another may lead to ageneralization of modeling behavior. As Bandura and Walters put it(1963:5): "Most children develop a generalized habit of matching theresponses of successful models." They cannot help developing at thesame time a generalized habit of observing those who are successful,or indeed of observing others to discover whether they aresuccessful. The matching presupposes the observation.
If we did not accept the reality of model learning, we should behard put to it to account, as we shall try to do later, for theeffect of a man's behavior not only on the others with whom he is inimmediate contact but also on members of an audience, who take nopart in the social behavior themselves but only watch it.
In proposition (I) we stated the effect of the success of anaction in obtaining a reward on the probability that a person willrepeat it. In speaking of the reward, we assumed that the value tothe person of the result of his action was greater than zero--thatis, he was not indifferent to it nor did he find it actuallypunishing. But the proposition had nothing to say about howrewarding the person found it. This variable, the degree of reward,we shall now bring in and call it value. The value in questionis always that of a given unit of the reward, no matter how that unitbe defined, since, as we shall see, the values of successive unitsmay change. The gross effect of this variable upon behavior may beexpressed by the value proposition:
III. The more valuable to a person is the result of his action,the more likely he is to perform the action.
The variable, value, may take either positive or negativevalues (now used in the mathematical sense of the term). The resultsof a person's actions that have positive values for him we callrewards; the results that have negative ones, punishments. The zeropoint on the scale is where the person is indifferent to the resultof his action. The proposition implies that just as an increase inthe positive value of the reward makes it more likely that the personwill perform a particular act, so an increase in the negative valueof the punishment makes it less likely that he will do so. And by anobvious extension of the stimulus proposition, if the occurrence of aparticular stimulus was the occasion on which an action was punished,the recurrence of the stimulus on a new occasion makes it less likelythat the person will perform the action. All of this is obviousenough.
Any action that has the result of allowing a person to avoid orescape punishment is rewarded by that result, and the person becomesmore likely to perform the action. Thus there are two classes ofreward: intrinsic reward and the avoidance of punishment. Similarly,there are two classes of punishrnent: intrinsic punishment and thewithholding of reward.
The use of punishment is an inefficient means of getting anotherperson to change his behavior: it may work but it seldom works well.On the other hand, it may give great emotional satisfaction to theman who does the punishing, and that is something not altogether tobe despised. Punishment may be enough when all that is required isthat the person stop doing something. Even then, if his action hasotherwise brought him valuable reward, it will soon reinstate itselfunless the punishment is often repeated and severe. Much moreefficient as a means of eliminating an undesirable activity is simplyto let it go unrewarded and thus eventually become extinguished, butapplying this method sometimes takes strong nerves. Suppose we wishto stop a child's crying when we suspect he cries only because itgets him attention. The best thing for us to do would be to ignorehim when he cried. But a mother often finds it heart-rending to carryout a policy like this. What if there really were something wrongwith him?
Punishment or its threat it still less efficient when it is usednot just to stop a person from doing something but to get him toperform a particular action. Then we punish him if he does notperform the action. The difficulty here is that punishment makesrewarding any action that allows him to avoid or escape thepunishment and not just the one we have in mind. Accordingly, we mustalso be prepared to punish or otherwise block off all avenues ofescape except that one. Doing so is apt to prove a costly business,especially if we add the cost of surveillance to determine whether heis really doing what we wish him to do. Punishment, moreover, is aptto produce hostile emotional behavior in the person punished, and wemust be prepared to cope with it. We shall have more to say aboutsuch emotions in later chapters. To get a man to perform an action byrewarding him if he does it, rather than by punishing him if he doesnot do it, avoids these costs-- but then the positive rewards may notbe available. We must face the fact that positive rewards are alwaysin short supply. Accordingly, while recognizing its disadvantages,there are times when we shall use punishment, for lack of anythingbetter, as a means of controlling behavior.
The things that men find rewarding-- their values-- are infinitelyvaried. Some of them are innate that is, genetically determined andtherefore shared by many men, such as the value set on food andshelter. Even some social values may be innate. It now looks as ifmen had evolved from apes that hunted in packs in open country. As aresult, we seem today to be more "social" in our behavior than ourcousins, the present anthropoid apes, just as wolves, who hunt inpacks, are more "social" than their cousins, the jackals. Men couldhardly have maintained pack behavior if they did not find social lifeas such innately rewarding. But this is speculation, and in anyevent, the capacity to find reward in social interaction must behighly generalized, not tied down to specific kinds of social reward.
What makes values infinitely varied is that, besides being born inmen and animals, they can also be learned. A value is learned bybeing linked with an action that is successful in obtaining a moreprimordial value. (See especially Staats and Staats, 1963: 48-54.)Suppose a mother often hugs her child and getting hugged is probablyan innate value in circumstances in which the child has behaveddifferently from other children and, as the mother says, "better."Then "behaving better" than others is a means to a rewarding end andis apt to become, as we say, "rewarding in itself." In other words,it is an acquired value. The reward may generalize, and thechild may be well on the way to setting a high value on status of allkinds. By such processes of linking, men may learn and maintain longchains of behavior leading to some ultimate reward. Indeed apart fromobvious anatomical differences and the use of language, the chiefdifference between the behavior of men and that of other animals maylie in the capacity of men to maintain longer chains relating, as wesay, means to ends. For the animals the ultimate reward cannot longbe postponed, if the sequence of behavior is not to fall to pieces.And even for men the ultimate reward must come sooner or later. Notethat the process by which values are acquired and linked to oneanother is the same for men as for other animals, but the number oflinks that can be put together in a chain is greater for the formerthan for the latter. As usual, the differences are not differences inkind but in degree.
Since different individuals may encounter different circumstancesin the course of their upbringing and thus acquire different values,men are apt to be more unlike one another in their acquired than intheir innate values. Yet there are some values that men in particularkinds of society would have difficulty in not acquiring. These arethe so-called generalized values, good examples of which aremoney and social approval. The act of fishing can be made moreprobable only by its success in getting a rather specific reward,that is, by the catching of fish-- though some men seem to usefishing only as an excuse for daydreaming or admiring the scenery.But money and social approval can serve as rewards for a wide varietyof actions and not just for some single kind. It is for this reasonthat they are called generalized values. In this book we shall beespecially interested in social approval as a generalized value.
So numerous are the values men can acquire, and so varied are thecircumstances in which they acquire them, that it is idle to make anygeneral statement about which ones they will hold. But if theparticular values held by particular individuals in particularcircumstances are known or can be reasonably inferred from theattendant circumstances (if the values in question can be taken asgiven and they often can be) then this variable can certainly be usedin accordance with our propositions to predict or account for otheraspects of behavior. But let no one talk about human values inabstraction from the past history and present circumstances ofparticular men.
In spite of all this talk about rewards, the reader should neverassume that our theory is a hedonistic one, concerned only withmaterialistic values. The values a man acquires may perfectly well bealtruistic. All our theory asks is that the values in question be aman's own values, not those that somebody else thinks he ought tohave. A man's success in obtaining altruistic values has just thesame effect on his behavior as his success in obtaining egotisticalones: he becomes more likely to perform the actions that have provedsuccessful, whatever they may be. My sisters and I once knew a womanwho set a high value on doing good to others, including ourselves.People sometimes say that virtue like hers is its own reward, that noexternal reward, no change in the behavior of others, is needed tomaintain it. We soon discovered that this was not true in her case.Her high-minded behavior did require an external reward, and it wasnothing less than our willingness to allow her to do good to us.Strangely enough, we were sometimes unwilling, and then she got asangry as the most materialistic of women deprived of the mostmaterial of goods. The language she used was more likely to disguiseher anger, but we soon became aware that it was anger just the same.We suspect that the same sort of thing may be true of other personswho hold altruistic values, which does not in the least mean that wemust be cynical about them or admire them less. They are out to helpothers, and why should the fact that in so doing they also rewardthemselves be held against them?
It is something to know what a man finds rewarding or punishingwhat his values, positive or negative, are. But the valueproposition (III) is not really concerned with what a person's valuesare. It is concerned rather with how valuable they are,how valuable a person finds a particular reward in comparison withother rewards. This question in turn must be divided into twoseparate ones. First: Is the same kind of reward more valuable on oneoccasion than on a different occasion? Does a person find catchingfish more rewarding, for instance, this morning than he will thisafternoon? Second: Is one kind of reward more valuable than adifferent kind on the same occasion? Does, for instance, a personthis afternoon find catching fish more rewarding than the results ofworking in his garden?
What we shall call the deprivation-satiation propositiondeals only with the first question. We may state it as follows:
IV. The more often in the recent past a person has received aparticular reward, the less valuable any further unit of that rewardbecomes for him.
If a man has received the reward often, he is beginning, as wesay, to be satiated with it. Its value for him decreases, and by thevalue proposition (III), he becomes less apt to perform an actionthat is followed by this reward. The proposition emphasizes the"recent past" because there are many rewards with which a man canonly temporarily be satiated. Food is the best example. If, on theother hand, a man has learned to value a particular kind of rewardbut has received it only rarely in the recent past, we say he isdeprived of it. For him, its value increases and by the valueproposition he becomes more apt to perform an action that is followedby this reward.
Obviously the deprivation-satiation proposition is not veryprecise and states only a very general tendency. What constitutes therecent past within which deprivation or satiation takes place must bedifferent for different kinds of rewards. Food can satiate menquickly, but it soon recovers its value. Most persons are not soeasily satiated with money or status, if indeed they ever can bewholly satiated. The reason is that these are generalized rewards,which can be used to obtain a large number of more specific ones.Unless a person is satiated with all the things that money can buy,he will not be satiated with money itself.
As for the second question--whether one kind of reward is morevaluable than a different kind on the same occasion--we can state nogeneral proposition that will help us to answer it. We can only tryto deal with particular cases. The number of possible comparisons isinfinite, and in each case we must rely as best we can on theaccumulated experience and knowledge that men have of other men andeven at times of our own knowledge of particular persons. We know,for instance, that a man caught out in a chill rain without a coat islikely for the moment to set a relatively high value on sheltercompared to other rewards, but that even then he would stay out inthe rain if seeking shelter meant losing his life. Again, we knowthat a man who is new to his job is likely to set a relatively highvalue on getting good advice on how to do it. At the other extreme,there are preferences, differences in value, which are far fromobvious. Thus we are told that the Chinese, faced with a choice ofdrinks, do not like milk and far prefer tea. It is easy to say thatthey have been taught to like tea and not milk, but that is notreally an answer to the question. Why should they have been taughtthe preference? The ultimate answer may lie in differences betweenthe traditional agriculture of China and that of the West. Yet if wehave some confidence that we know what the values of a man or of agroup of men are, even if we do not know why they hold thesevalues--if we can take their values as given in givencircumstances--we can, with the help of our general propositions,make some good bets on what their other behavior is apt to be.
We have begun by separating the two kinds of questions aboutrelative values. In the end we may have to bring them together again,so that we do not leave the different values of the same reward atdifferent times wholly unrelated to the different values of differentrewards at the same time. In general, a man's satiation with aparticular reward renders all his other rewards relatively morevaluable to him. Moreover, it may turn out that the values can beplaced in some kind of rank order, or hierarchy, of values, such thatunless a man is first satiated with a particular kind of reward, thenext higher kind in the hierarchy will have little value for him. Orrather, to bring in the success proposition (I), the man, if notactually satiated, must be pretty sure of getting enough of the firstkind before he can set much store by the next. (See Maslow, 1954.)Thus unless a man knows where his next meal is coming from, he isunlikely to set a high value on some other reward such as status: hecan forgo status more easily than he can food. Americans are said toset a high value on democratic processes. Would they do so ifdemocracy got in the way of their getting enough to eat? One mayguess that democracy would be the loser, but fortunately mostAmericans have not had to make the choice. There are some intangibleand ideal rewards on which men will set a very high value but only ifother "lower" needs are being met. We can only raise the questionhere. We know too little about the ways in which men rank values in ahierarchy of this sort.
So far we have had nothing to say about the emotional behavior ofmen, and thus have left out much that makes them humans. A fullerpsychology than ours pretends to be would include severalpropositions about emotional behavior, among the most important ofwhich would be statements about the causes and effects of anxiety.But in this book, in order to keep the treatment as simple aspossible, we shall introduce only one proposition about emotionalbehavior, the only one we shall badly need in order to explain thefindings about social behavior considered in later chapters. Thisproposition we call the aggression-approval proposition and wecan perhaps state it most conveniently by dividing it into two parts,one concerned with aggression and the other with approval.
The first part is usually called the frustration-aggressionhypothesis (Miller and Dollard, 1941):
Va. When a person's action does not receive the reward heexpected, or receives punishment he did not expect, he will be angry;he becomes more likely to perform aggressive behavior, and theresults of such behavior become more valuable to him.
Let us now comment on each clause of this complicated proposition.
When a person does not get what he expected, he is said to befrustrated. A purist in behaviorism would not refer to theexpectation at all, because the word seems to refer, like other wordssuch as "purpose," to a state of mind. Yet if we did not use it wecould only replace it by a long circumlocution, without anyoffsetting gain in rigor. Nor need the word refer only to an internalstate; it can refer to wholly external events, observable inprinciple not just by the person himself but by outsiders. What a manexpects to get by way of reward or punishment under a given set ofcircumstances (stimuli) is what he has in fact received, observed, orwas told others received, under similar circumstances in the past;and none of these things are private events confined within theindividual's head. This is what we shall mean by the wordexpectation. In later chapters we shall be especiallyinterested in a kind of experience many men have shared, which givesrise to a very generalized expectation Aristotle (1967) called a ruleof distributive justice. Since what a man expects depends in the longrun on what has actually happened to him, his failure to receive anexpected reward, if repeated often enough, finally results in achange in what he expects. What was once unexpected now becomes theexpected thing, and his anger fades, but it may take a long time todo so.
When a man is frustrated, he is apt to feel some degree of theemotion we call anger. Again, a purist in behaviorism might not referto anger in his version of the proposition but only to the aggressivebehavior; we keep the anger in so that we may not do too greatviolence to the common sense of men. Men show that the experience ofanger has much the same meaning for all of them through the ease withwhich they can communicate to others the fact that they are angry. Nodoubt the more valuable to a person is the reward he expected or themore painful the punishment he did not expect, the greater is hisfrustration and hence his anger.
When a man is frustrated, he is apt to perform aggressive actions.These are actions that attack, break, hurt, or threaten the source ofthe frustration, whether the real source or what the man perceives itto be. If for any reason the real source cannot be attacked, almostany target will do in a pinch. The target may of course be aninanimate object. We do not kick a stuck door just because a kickwill help to open it, for it usually will not. We kick the door inorder to hurt it. But in this book we are naturally much moreinterested in human sources of frustration and targets of aggression.In anger, moreover, the successful results of aggressive actionreward a man as they would never have done without the anger. When weare furious at someone and hit him, the sight of his wincing underour blow becomes intensely rewarding.
In our first four propositions we were dealing with voluntary or,as the behaviorists call it, operant behavior. Operant behavior andemotional behavior such as aggression differ in the initialconditions that make their appearance more probable. No previousstimulus can automatically get a man to perform an operant the firsttime. He must just happen to perform it, even as a matter of chance,and be rewarded by it before he will perform it again. Only after hehad been rewarded will the attendant stimuli begin to get somecontrol over his action. Aggressive behavior can, on the contrary, beautomatically produced the first time by a stimulus-- the failure ofan action to get the expected reward. In this respect, its initialrelease by a stimulus, aggression resembles a reflex like thefamiliar knee jerk.
Yet in another and more important respect aggression differs froma reflex. A reflex cannot be learned--one cannot learn to do aconvincing knee jerk but aggression can be learned. That is, anaggressive action, originally purely emotional, can become voluntary.Whatever the conditions of frustration that led a man to perform anaggressive action in the first place, if in fact his aggression isfollowed by a reward, wholly apart from the satisfaction of hisanger, he becomes more likely to perform it again, just as if it werean ordinary operant. As we all know to our cost, aggression may pay,and if it does will be repeated. Many men and groups use aggressionsimply as an instrument for attaining practical results. In theircase, the aggression may create the anger, not the anger, theaggression. But by the same token, a man may come to performaggressive actions less often if they have not been successful orhave actually resulted in punishment. He may learn to get his outwardaggression, if not his inward anger, under control. Or he may stillattack but may learn to displace his attacks from targets thatrespond with punishing reprisals to less dangerous ones. In this bookaggression and the like will be treated as if they were, at one andthe same time, both emotional and voluntary activities.
Let us now turn to the second part of the aggression-approvalproposition. We have long believed that the special emphasispsychologists have placed on the first part, that is, thefrustration-aggression hypothesis, has tended to give a one-sidedview of the emotional behavior of men since it has pointed only totheir negative emotions. But if they can be frustrated and hate, theycan also be fortunate and love. Let us therefore propose with somediffidence the following as the second part of theaggression-approval proposition:
Vb: When a person's action receives reward he expected,especially a greater reward than he expected, or does not receivepunishment he expected, he will be pleased; he becomes more likely toperform approving behavior, and the results of such behavior becomemore valuable to him.
If the reader is not altogether happy with the wordspleased and approving which we have used in theproposition--and we confess we are not altogether happyourselves--let him find his own opposites to angry andaggressive.
Many of the comments that we made about the first part of theproposition we may repeat, mutatis mutandis, for the secondpart. Two points are of particular importance. First, though menoften give what we call their spontaneous admiration to others whohave provided them with unusual reward, they obviously can also learnto give approval to others simply as an instrument for gettingfurther reward from them, apart from the expression of the admirationitself. Approval can become such an instrument because many men findthe approval they receive from others rewarding, just as they findaggression punishing. In short, approval like aggression may become avoluntary as well as an emotional action. In later chapters we shallhave a great deal to say about approval as one of the most importantrewards of social behavior. Second, if what was once an unexpectedand unusual reward becomes by repetition an expected and usual one,the person's original emotional reaction will tend to decline instrength, which need not mean that he will cease to use approvalinstrumentally. Following The Human Group (Homans, 1950) weshall refer to variables like anger and approval, whetherspontaneously or instrumentally expressed, as sentiments.
The Propositions as a System of Propositions
Now that we have stated the general propositions we shall lateruse in explaining social behavior, we must make one or two commentson the set of propositions as a whole. We have stated eachproposition baldly, without qualifications, without adding the escapeclause that each holds good only under the condition that "otherthings are equal." The reason we have done so is that what these"other things" are and where they are "equal" are determined for eachproposition by the other propositions in the set. The effects thatwould be predicted by any one of the propositions may, in concretecases, be masked or modified by the effects of other propositions inthe set. That is, the set must be taken as a whole system ofpropositions.
Let us offer just one crude illustration. The success proposition(I) says that the more often an action is rewarded, the more often aman will perform it. But this relationship certainly does not alwayshold good in real life. For if the reward comes often enough, thevalue the man sets on a further unit of it will, by the satiationproposition (IV), decline, perhaps even to the extent that he isindifferent to it for the time being. But as the value of the rewarddecreases, then the man, according to the value proposition (III),becomes less likely to perform it and not more likely. What followsfrom the three propositions taken together is that a man will performan action at the fastest rate when the action is rewarded only justoften enough to keep him slightly deprived of it. If he were whollydeprived, it would mean that his action was utterly unsuccessful ingetting the reward; and complete lack of success leads to inactionjust as much as satiation does.
The propositions imply, if we did not know it already, that thepast history of men makes a big difference to their present behavior,and not just the recent past but often the past of long ago. A man'spast history of success, of stimulation, of the acquisition of valuesall affect the way he behaves now. The choices he made in the pastmay still be limiting the opportunities available to him today, or hemay perceive them as limiting; hence the great weight attached to aman's early experience by all schools of modern psychology. The illeffects of some early experiences may of course be overcome, but itmay be difficult to do so there is something to be overcome.
The effect of past experience extends beyond the history ofindividuals to the history of societies. Since children learn much oftheir behavior and values from parents and other members of the oldergeneration, the past culture of a society tends to perpetuate itself.We need not believe that a society maintains itself by teaching itsmembers just those actions it is prepared to reward, just thosevalues it is prepared to satisfy. If a child acquires at his mother'sknee a value like independence, he may, when he grows up, try tochange his society radically instead of preserving it. Indeed we knowthat old values and actions whose success in attaining these valueswere learned long ago may, in new circumstances, lead to radical,unforeseen, and quite unintended social change. For instance, it wasold values and old types of action that finally, in newcircumstances, created the Industrial Revolution. Yet there is alwayssome tendency for past behavior to maintain itself, at least in thesense that every new generation has to start from something thatalready exists; it can never make a wholly fresh start. Indeed themen of the past may, in pursuit of the values of their time and byits methods, have created institutions to which their descendants arecommitted, at least to the extent that they cannot change all theirinstitutions at once. Past institutional commitments have the sameeffect on the history of societies that past choices often have onthe history of individuals.
It is this historicity that makes it difficult to explain humanbehavior and human institutions. Other sciences do not suffer from itnearly as much. It makes little difference in explaining themechanical action of a lever what sort of past history the materialit is made of may have had. All it needs to be is strong enough, andthere is an infinite number of paths by which it could have becomestrong. Were there only a single path by which a lever could meet therequirement, we might be called upon to explain why a particularlever took just that path and not some other. But there are cases,such as that of magnetic hysteresis--the tendency of a piece of ironto acquire a magnetic field in particular circumstances, which itthen tends to preserve in new circumstances in which historicity doesmake a difference even in physical science, and then the science hasas much trouble in dealing with it as social science does.
Historicity makes least trouble when the forces acting on men orsocieties are convergent, when, that is, strong forces are tending tomake men or societies more like one another, whatever their initialdifferences. In explanation we can then afford to neglect the detailsof the paths by which they reached this similarity. The realdifficulty comes when we deal with divergent phenomena, such as theone we cite in the proverb: For want of a nail the shoe was lost; forwant of a shoe the horse was lost; and so on. Then a force weak initself but just tipping the scales in a balance of stronger forcesmay have great and widening effects over time.
If the present precipitates of a man's past history--the acts inwhich he has been successful, the stimuli that have accompanied them,the values he has acquired--can be taken as known and given, we canthen apply our general propositions to these given conditions inorder to explain or predict his behavior. But we seldom know enoughabout these conditions to predict or explain accurately. In practice,we cope with the problem as best we can by predicting or explainingthe behavior of a number of men at a time, so that if we go wrong insome instances the general tenor of our results may still hold good.We cope with it also by assuming that, the more similar are the known"backgrounds" of men, the more similar their past histories must havebeen, and so the more similar their behavior in present circumstancesis likely to be. We assume, for instance, that sophomores in Americanuniversities are apt to have shared some kinds of past experience. Inthis respect they resemble one another more than they do, say, theirFrench contemporaries. If we know from acquaintance with Americansophomores what these similarities are, if we can take them as knownand given, we can often predict with the help of our generalpropositions what some other features of their behavior are likely tobe; but we shall never be able to do so perfectly. Through someaccident in his upbringing a particular sophomore may, for instance,have acquired values altogether unlike those of men who appear on thesurface to share the same background, and so may not respond in thesame way to our questionnaires or experimental manipulations. Forthis reason our findings, our correlations, can at best be onlystatistically significant and never perfect.
In any event, this book cannot include a treatise on what iscalled developmental psychology, the study of the process by which anewborn child becomes morally and intellectually an adult. Instudying social behavior we shall simply assume that we are dealingwith persons who have experienced some of the same gross features ofpsychological development.
This may be a useful point at which to enter a warning. Althoughthe general propositions we shall use are often called thepropositions of "learning theory," we are far from believing that menare equally likely to learn anything in the way of behavior, providedonly that they encounter in the social and physical environment theappropriate stimuli and rewards. They do not, so to speak, start lifeas blank sheets of paper on which the environment can readily writewhatever occurs to it to write in the way of learning. Not only theirexperience but their genetic endowment--not only nurture but nature,to use the neat antithesis--determines what they learn.
It is not a question of which is the more important, nurture ornature, though that was the question psychologists asked themselvesfor several decades. The real question is how the experience of meninteracts with their genetic inheritance. To take an obvious example,a big, strong man is more likely than a small, puny one to find thatcertain kinds of action, such as physical aggression, get him rewardin his contacts with other men. Accordingly big, strong men are morelikely than small, puny ones to learn and to perform physicallyaggressive actions. Size and strength, apart from differences innutrition, are genetically inherited characteristics. Yet they do notaffect the behavior of men directly; they affect it only because theyalter the contingencies in which learning takes place. Lessobviously, an intelligent person may be able to learn, underappropriate conditions of stimulation and reward, types of behaviorthat a less intelligent person cannot learn, or cannot learn soquickly. Yet intelligence, if it is the sort of thing measured byintelligence tests, is certainly in some degree geneticallyinherited. Mankind itself, as we have already suggested, has probablyinherited and not just learned a generalized capacity to find socialcontact with other human beings rewarding, and if it had notinherited this value, it would probably not be able to learn andmaintain some of the forms of social behavior it does in fact learnand maintain. It may also turn out that individuals inherit differentdegrees of this general value, and that these differences,interacting with their experiences of the social environment, producefurther differences in learning.
We have less need to bring in past history in explaining somekinds of behavior than others. In order to understand this point letus begin by looking at a proposition that in effect sums up the firstthree of our propositions, those concerned with success, stimuli, andvalue. (As for the last two propositions, thedeprivation-satiation proposition (IV) states one of thecauses for a change in the value of a reward and theaggression-approval proposition (V) states the conditionsunder which the results of certain kinds of actions become valuable.)The proposition has been called the principle of rational choice orthe rationality proposition. (See, for instance, Harsanyi,1967.) It may be stated as follows:
In choosing between alternative actions, a person will choosethat one for which, as perceived by him at the time, the value, V, ofthe result, multiplied by the probability, p, of getting the result,is the greater.
In so acting, a person is said to maximize his expected utility(Ofshe and Ofshe, 1970: 3). In the case of repeated actions, therationality proposition implies that, if the rate at which a manperforms an action is designated by A, then
A = pV (10)
Suppose, for example, that a man faces a choice between twoactions. The first, if successful, will bring him a result worth, letus say, three units of value to him, but he estimates the chance thathis action will be successful as only one out of four. The secondwill bring a result worth only two units, but he estimates its chanceof success as one out of two. Since 3 x 1/4 is less than 2 x l/2, therationality proposition predicts that the man will take the secondaction.
The proposition becomes simplified if each of the alternativeactions is certain of success, so that p = 1. Then the man's choicedepends only on the relative value of the results. This condition is,of course, sometimes satisfied in real life. For instance, in thekind of market which classical economics for its purposes of analysisassumed to exist and which some real markets in fact resembled, awilling buyer could always find a willing seller. His success in theact of buying was certain, and so his decision to buy depended onlyon whether he was ready to pay the price-- whether the value of thegoods bought was greater to him than that of the forgone alternativeof keeping his money to spend on something else. Much economics usesthe rationality proposition in this stripped-down form as its singlegeneral proposition, often without actually stating it.
Now let us consider in what sense the rationality propositionembodies, or corresponds to, our first three propositions. It statesthat one of the two factors determining whether a man will perform anaction is its probability of success as he perceives it. But theproposition says nothing about what in turn determines hisperception. In the case of actions repeated over time, one of thedeterminants of his perception will be the actual frequency withwhich the past action has been followed by the reward. Then thisaspect of the rationality proposition is stated by the successproposition (I): the more often the action is in fact rewarded, themore often it is performed. Let us write this in the form ofHerrnstein's equation (1, p. 21), substituting for his variableP, the rate at which a pigeon pecks, the more general variableA, the rate at which a man performs any given action.
A = kR (11)
Here R, the actual frequency with which an action has beenrewarded replaces the variable p, the perceived probability ofsuccess, in the rationality proposition.
The rationality proposition also embodies our stimulus proposition(II), in that a man's perception of the probability that his presentaction will be successful is further determined by the similarity ofthe circumstances attending the action at present to those underwhich the action was successful in the past.
Now let us look at the constant of proportionality, k, inHerrnstein's equation. It is a constant there because the actions ofhis pigeons were identical in form (pecks) and in reward (grain).They differed only in the rates, R, at which pecks at twodifferent keys were rewarded. But most alternative actions differ inform and in the kind of reward they receive. When they differ inform, they generally also differ in cost; and when they differ inreward, they generally also differ in the value of the reward. Thatis, they differ in the net value of the reward they receive. In amore general formulation the constant in Herrnstein's equation mustbe replaced by the variable, V, and then the equation takesthe form
A =RV (12)
We should now be able to recognize how each variable in thisformulation corresponds to a variable in the rationality proposition,remembering that "corresponds to" does not have the same meaning as"is identical with."
Just as the rationality proposition has nothing to say about why aman perceives his chances of success as high or low, so it hasnothing to say about why he has acquired certain values and notothers, why he sets a higher value on one reward than he does onanother, or why he values a particular reward more highly on oneoccasion than he does on another. A more fully developed psychology,and one that deals with the effects of a man's past history on hispresent behavior, is needed to account for these things.
Yet we must always remember that, besides what a man perceives tobe his chances of success in various actions, there are always theactual chances of success as given by the outside world independentof his perceptions, including at the extreme the certainty that somekinds of actions will be successful. Suppose we have reason tobelieve that a man's perceptions of the outside world are accurate,that the outside world and his map of it coincide. Suppose we canassume further that his values are common values, values that mostpersons share, or that are common to a class of men to which he isknown to belong. Then, in explaining his behavior, we can neglect thedetails of his past history. For example, if we know a man is askilled carpenter, then we explain what he does in building a house,though not necessarily in other activities of life, by the laws ofphysics, so to speak, and not by reference to his individual past.Nor do we have to account for his particular values. We confidentlyassume that, simply because he is a carpenter, he will set some valueon building a house that will stand up. Apart from professionalpride, he will set some value on it if only because he will not earnhis living very long as a carpenter if his houses will not stand up.If a man's chances of success in his actions are given by the outsideworld, and if his values are those known to be held by the class ofmen to which he belongs--if, that is, his values and his perceivedchances of success can simply be taken as given--then the rationalityprinciple alone, without the use of a broader set of propositions,will explain and predict his behavior pretty well. Of course, ourcarpenter may make mistakes, but in the long run his behavior willapproximate to what the rationality proposition predicts. Much humanbehavior, of course, meets these conditions and can be explained inthis way
But much behavior cannot be explained by the rationalityproposition. If a man's values are somewhat queer, not as easilytaken for granted as those of a skilled carpenter in the exercise ofhis trade, and if his chances of success are not given by the outsideworld, or at least not accurately known by him, then the rationalityproposition by itself may not help us much. Our favorite examplesamong the enormous number that could be cited, is the decision byWilliam the Conqueror (then simply the Bastard), Duke of Normandy, toinvade England in 1066. That he set a high value on the result ofsuccessful conquest, becoming king of England, those of us who havesome knowledge of other feudal lords will find no difficulty inaccepting. But what about the other term in the rationalityproposition, his chance of success as perceived by him? He could notknow what his chances were. As the economists would say, hiscondition was one of uncertainty and not of risk. In risk the odds onsuccess are accurately known; in uncertainty they are not known. Evenif he were fairly sure he could get an army and a fleet together,there remained for him the dangers of a sea voyage, of landing on ahostile shore, and of battle with an English army under theexperienced and hitherto successful command of Harold Godwinsson.Defeat in battle would almost certainly mean death. William'scontemporaries might well have judged his chance of success to besmall. On the record, we have no reason whatever to believe that hewas a foolish man. Why then did he go ahead with the enterprise? Intrying to answer this question it is surely relevant to point to thealmost unbroken series of his military victories over the precedingtwenty years. It is not the rationality proposition but the successproposition that will account for the effect of these victories onhis decision. Past success in military action made his futuremilitary action more probable. Or, as we say in ordinary language,past success had given him confidence.
Certainly, if we keep the rationality preposition firmly in mind,we shall never forget that human action is determined by two kinds offactor, not one. Many persons, including many social scientists, talkas if what determines a man's action was his "motivation" alone-- inour terms, the value he sets on the result of his action. But a manmay be highly motivated in this sense and still not take action, ifhis similar actions in the past have been uniformly unsuccessful.Again, some social scientists talk as if the reason why somelower-class groups, like blacks in the United States, remainunassimilated to the larger society in which they live is that theirvalues are different from those of other groups in the society. Theirvalues may be just the same as those of the rest of society, but iftheir actions have been, for whatever reason, unsuccessful inobtaining those values, they will turn to alternative actions. Ifthese alternative actions are successful in obtaining a differentkind of reward, their actions may keep them as effectively cut offfrom the rest of society as if their values had been different allalong. The rationality proposition serves to remind us that action isdetermined by success and value jointly
We must recognize that not one of the propositions we have putforward is new. It is not just that they are not original with us buthave been put forward, if not in the same words at least with thesame substance, by many psychologists. They were not new even to thepsychologists. Undoubtedly in one crude form or another they havebeen known to men for ages. They hardly come as a surprise to any ofus, though some of their further implications, in psychopathology forinstance, may indeed be surprising. Accordingly, the propositions mayappear to be obvious, but they are not to be rejected just for thatreason. It is also obvious that the obvious need not be untrue, andwhat we seek is the truth. Perhaps there are general propositions insocial science that remain to be discovered, but those that have longbeen known are not made trivial by that fact.
Let not a reader reject our argument out of hand because he doesnot care for its horrid profit-seeking implications. Let him askhimself whether he and mankind have ever been able to advance anygeneral explanation why men change or fail to change their behaviorother than that, in their circumstances, they would, as they see it,be better off doing something else or that they are doing well enoughalready. On refiection, the reader will find that neither he normankind has ever been able to offer another explanation; it is atruism. He may ease his conscience by remembering that if hedonistsbelieve men take profits only in materialistic values, we are nothedonists here. If men's values are altruistic, they can take aprofit in altruism too. In fact, some of the greatest profiteers weknow are altruists.
Though we have called the last proposition the rationalityproposition, we should be aware of the different meanings men havegiven to the word rationality. The first is rationality fromthe point of view of some omniscient outside observer who viewsbehavior as irrational if he knows its reward is harmful to a man. Aperson is irrational if he pursues rewards that he ought not to findvaluable. In this sense a man who takes certain kinds of drugs,including tobacco, behaves irrationally. So is a masochist, a man whofinds punishment rewarding, though here the criterion of "goodness"or "health" is not so clear. But since in this book we are notinterested in how men ought to behave but only in how they do in factbehave, we care not here--though surely we do care elsewhere--whether a man's values are rational or not. What we are interested inis what he does with them once he has somehow picked them up. Supposea man is a masochist. Is it still true that, if he has taken a lot ofpunishment recently, he will find further punishment for the momentsomewhat less valuable? Will the first kick in the teeth give himmore of a kick than the last? Will the satiation proposition apply tohim too? These are the types of question we are interested in.
Even if, in the eyes of the omniscient observer, a man's valuesare rational, his actions may still be irrational if they are notwell designed to obtain these values. A man may be ill-informed ormisperceive the situation that faces him or fail to realize that adifferent action might be more successful or successful at a lowercost. If the standard of rationality is set by the omniscient outsideobserver, the man's behavior is then irrational. All we can say isthat the rationality assumed by the rationality proposition is not ofthat sort. Whatever a man's information, perceptions, and designsmight conceivably be, if he does not in fact possess the bestpossible ones but acts in accordance with those he does possess,though they may be wrong or inadequate, he is acting rationally.
The second meaning of rationality is closely related to the first.Whatever a man's values may be, his behavior is irrational if it isnot consciously calculated to get him the largest supply of thesevalues in the long run. Here the emphasis is not on the kind of valuebeing pursued-- it may be capital gains or eternal salvation-- but onthe way it is being pursued: the emphasis is on calculation and thelong run, the longer the better. By this standard an irrational manis one who is either unwilling to forgo some immediate reward inorder to invest in some greater future or unwilling to acquire theknowledge and make the calculations that would show him how to reachthat future. A large part of many sciences, from divinity to theTheory of Games (Von Neumann and Morgenstern, 1944), is devoted toproviding him with this knowledge and enabling him to make thecalculations. The Theory of Games should, for instance, make himbetter able to choose a strategy among alternative courses of action,when the risks and returns of each are matters of probability, notcertainty.
Although calculation for the long run plays its part in humanaffairs, we make no special allowance for it in our propositions.Above all we assume that the propositions hold good whether thebehavior in question is conscious or unconscious. In the fields ofhuman endeavor we shall be interested in, conscious and unconsciousbehavior very often come out at the same place. One man may offeranother a greater reward in social approval in return for a morevaluable kind of help; his behavior may be utterly economic, withouthis being any more conscious of what he is doing than a pigeon is--but then we do not know how conscious a pigeon can be.
We neither rule out nor rule in conscious calculation. Our firstjustification is that we shall not often need it in order to explainthe research results considered in this book. And our second lies inplain sight: calculation for the long run is the exception and notthe rule. The Theory of Games may be good advice for human behaviorbut a poor description of it.
The fact is that the question whether behavior is rational or notby any of the definitions proposed is irrelevant to our purposes. Allthe good advice, from ethics to economics, that wise men give theirfellows is meant to change behavior and not to explain it; but ourbusiness is with explanation. The advice tries to answer thequestion: Given that you value the attainment of certain ends, howcould you have acted so as to attain them more effectively? But whatmen or pigeons could have done is what they did do, and much socialscience has gone to show some of the surprising reasons why theycould not do in fact what some wiseacre says they could. This doesnot mean that all the advice goes for nothing. So far as men willtake it, so far as they will learn from it, it may change the waythey behave the next time. But behavior observed is behavior past,and for the purposes of explaining how men have indeed behaved it isseldom enough to ask if they were rational from the point of view ofan omniscient observer. The relevant question is what in factdetermined their behavior-- though of course the advice they havelistened to may be among the determinants.
The persons who will appear in this book are, if you like, no lessrational than pigeons. If it be rational of pigeons to learn and takethe shorter of two paths to a reward, so it is of our men. Theychoose among a few alternatives immediately open to them; they choosewith little regard for the really long-term results of their choice,which sometimes surprise them. But the short-term results they doknow, and they often know them less as matters of probability than ofcertainty. Within these limits, our men do not choose foolishly--that is, at random-- but only in the way our propositions say theydo. All that we impute to them in the way of rationality is that theyknow enough to come in out of the rain unless they enjoy gettingsoaked. To be sure, such rationality as we have now left them may notamount to much, for rational behavior in our sense, and not in thatof the omniscient observer, is only behavior that is determined.
Let us make sure we are not snobs about the common pigeon or thecommon man. When the future is not easily foreseen and science isweak, the pursuit of immediate reward is by no means irrational evenby the austere standards of the Theory of Games. "A bird in the handis worth two in the bush" is by no means always an unintelligentpolicy. And so far as the pursuit of rationality entails study,forethought, and calculation, and such things hurt, as they often do,the pursuit of rationality is itself irrational unless these costsare reckoned in the balance. The costs of rationality may makerationality itself irrational.
The rule we have found to be implicit in the behavior of theemployees of the Customers' Accounting Division we believe to beimplicit in the behavior of many persons in many different societies.Nor does it remain merely implicit in behavior; it is oftenexplicitly stated. Thus Jouvenel (1957: 149) writes: "What they [men]find just is to preserve between men as regards whatever is inquestion the same relative positions as exist between the same men asregards something else." But we ought to consider with special careand veneration the first statement about distributive justice, whichwe owe to Aristotle. It appears in Book V, Chapter 3, of hisso-called Nichomachean Ethics, which consists of drafts of hislectures on ethics, put together by himself or, more probably, by hispupils. He distinguishes between corrective justice, such as theproper punishment of criminals, and distributive justice, which iswhat we are interested in. Of it he writes (Aristotle 1 967: 269):
It follows therefore that justice involves at leastfour terms, namely two persons for whom it is just and two shareswhich are just. And there will be the same equality between theshares as between the persons, since the ratio between the shareswill be equal to the ratio between the persons; for if the personsare not equal, they will not have equal shares; it is when equalspossess or are allotted unequal shares, or persons not equal equalshares, that quarrels and complaints arise.
A little later he goes on to say:
Justice is therefore a sort of proportion; forproportion is not a property of numerical quantity only, but ofquantity in general, proportion being equality of ratios, andinvolving four terms at least.
Let us try to restate in our own terminology what Aristotle issaying here. Distributive justice involves a relationship between atleast four terms: two persons, P1 and P2, one of whom can be assessedas higher than, equal to, or lower than, the other; and their twoshares, or as we would say, rewards, R1 and R2. The condition ofdistributive justice is satisfied when the ratio of the measures ofthe persons is equal to the ratio of the measures of their respectiverewards. That is, if the two persons are equal, they should, injustice, receive equal rewards; if one is better than the other, heshould receive the larger reward. That is, the condition ofdistributive justice is satisfied when:
P1 / P2 = R1 / R2 (1 )
In the fourth century B.C., Aristotle was saying just what wasimplicit in the behavior of the employees of the Customers'Accounting Division, provided only we recognize that what he callsthe measure of a person should include both his investments and whathe actually contributes to social exchange.
We believe, though we confess we do so without adequate evidence,that something like the rule of distributive justice is eitherexplicitly stated or stands as an implicit major premise in thearguments and behavior of many men in many, and probably all, humansocieties. We do not argue in the least that it is the noblest ruleof distributive justice a philosopher or saint could devise; we argueonly that it represents what many men in fact find fair. No doubt,through their proverbs and, by implication, through their comments onthe behavior of themselves and others, members of the oldergeneration in many societies teach the young some form of the rule,as a condition that ought to obtain, but often of course does not.Though, again, we cannot claim to be experts on the development ofmoral standards in children, we also believe that, even if the rulewere not taught to the young, they would discover it anew forthemselves every generation.
The reason we think so is that the rule of distributive justice isa statement of what ought to be, and what people say ought to be isdetermined in the long run and with some lag by what they find infact to be the case. Now we have seen that ordinary exchanges betweenpeople, exchanges such as we have illustrated in our little group ofPerson, Other, and the Third Man, when they are unconstrained by fearof punishment or by the actions of outsiders, do in fact result inpersons who produce superior contributions to the exchange,contributions recognized as superior by the parties themselves,getting superior rewards; in persons who produce similar (equal)contributions getting similar rewards, and so on. Thus, as we havepointed out so often, Other in our little group gives the best adviceand gets the most approval. So many people, we believe, encounterthis process of distributing rewards so often in their own actualexperience that they come to generalize it as the kind ofdistribution that ought to obtain always.
Unfortunately, the fact that people accept the same general ruleof distributive justice need not mean that they will always agree onwhat is a fair distribution of reward between them. Even if theyconcede that reward should be proportional to investment andcontribution, they may still differ in their views of whatlegitimately constitutes investment, contribution, and reward, andhow persons and groups are to be ranked on these dimensions.Aristotle himself recognized the difficulty. Speaking at thebeginning of Book V of his Politics (Aristotle, 1967: 371-372)about the conflict between the parties of oligarchy and democracy inthe cities of Greece, he stated:
Thus democracy arose from men's thinking that if theyare equal in any respect they are equal absolutely (for they supposethat because they are all alike free they are equal absolutely);oligarchy arose from their assuming that if they are unequal asregards some one thing they are unequal wholly (for being unequal inproperty they assume that they are unequal absolutely) and then thedemocrats claim as being equal to participate in all things in equalshares, while the oligarchs as being unequal seek to have a largershare, for a larger share is unequal.
That is, the oligarchs and the democrats, while acceptingimplicitly the same general rule, differed about how it should beapplied. The oligarchs thought property the only legitimateinvestment. Since they had more property than the democrats, theybelieved they should have more than the democrats in other respects,such as political rights. The democrats, on the other hand,considered free birth the only legitimate investment. Since in thisrespect they were the equals of the oligarchs, they believed theyshould be equal to the oligarchs in political rights. (Presumablyboth parties agreed that in all respects they were superior to theslaves and should allow them no rights at all.)
What was true of ancient Greece has probably been true of allother groups and societies. In the Customers' Accounting Division,management talked as if only skilled work and good wages ought toenter into the balance of fair exchange, whereas the ledger clerksthought that other features of their job, such as its lack ofautonomy, ought to be brought in as well. In industry everyone agreesthat promotions ought to be in line with the employees' investmentsand contributions, but the general rule is not of much help inanswering the next question: Which should count most in gainingpromotion for a man: his skill or his seniority? Many labor unionshave in effect come down on the side of seniority, partly becausedifferences in seniority can be determined more objectively than candifferences in skill, and so leave less room for argument.
Moreover, what people agree on counting as legitimate investmentschange from time to time in many societies. Once upon a time ancientlineage counted as a legitimate investment, but in modern Americansociety it counts officially not at all, though one can sometimesfeel it playing its part unofficially. Once upon a time, too, thefact of being a man rather than a woman counted as more of aninvestment than it does today in America: a man doing the same job asa woman might demand more pay than hers. Today he is much less likelyto make such a claim. And the same, of course, was true of being awhite rather than a black.
All the arguments about surplus value from John Ball to Karl Marxand his followers amount to one long attempt to prove that whatemployers and owners count as their investments and contributionsought not to be so counted, and that therefore they get more thantheir fair share of the returns from economic enterprise and exploitthe workers. Naturally none of the arguments prove the point; suchthings are not capable of proof; they are matters, so to speak, oftaste. When someone says that one group is exploiting another, allhis words mean is that he personally does not approve of the way inwhich rewards are distributed between the two groups. They mean nomore: there are no objective means of determining whetherexploitation is taking place.
Only perhaps for rather brief spans of time or in rather smallgroups are men fully agreed not only on what the rule of distributivejustice is but also on what particular investments, contributions,and rewards should fairly be placed in the scales and at whatweights. Men certainly assess their own investments, contributions,and rewards; but to make a rule of justice work, they must assessthose of others by the same measures and on the same dimensions. Theothers, for their part, must agree at every point, which does notmake consensus any easier to achieve. The evidence, as we mightexpect, is that it is easier to achieve among persons who, throughsimilar experience, reflected in similar backgrounds, have acquiredsimilar values; but even here consensus is always in danger ofbreaking down. This means that there is no just society, though theremay be societies and groups that are more or less just, to the extentthat their members agree in their evaluations and maintain a roughproportionality between contributions and rewards. By these standardsthe Customers' Accounting Division was a pretty just place. Exceptfor the ledger clerks everyone believed that she was gettingsomething like her fair share relative to the others, and as for whatwas happening to the ledger clerks, almost everyone agreed at leastthat it was an injustice-- which is more than can be said aboutinjustice in many places.