Canadian Journal of Philosophy - Web Hosting at UMass ... Flipbook PDF

Canadian Journal of Philosophy Methodological Individualism and Marx: Some Remarks on Jon Elster, Game Theory, and Other

---

326 Views
117 Downloads
FLIP PDF 2.38MB

DOWNLOAD FLIP

REPORT DMCA

Canadian Journal of Philosophy

Methodological Individualism and Marx: Some Remarks on Jon Elster, Game Theory, and Other Things Author(s): Robert Paul Wolff Source: Canadian Journal of Philosophy, Vol. 20, No. 4 (Dec., 1990), pp. 469-486 Published by: Canadian Journal of Philosophy Stable URL: http://www.jstor.org/stable/40231710 Accessed: 25/05/2010 11:11 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=cjp. 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 [email protected].

Canadian Journal of Philosophy is collaborating with JSTOR to digitize, preserve and extend access to Canadian Journal of Philosophy.

http://www.jstor.org

CANADIAN JOURNAL OF PHILOSOPHY Volume 20, Number 4, December 1990, pp. 469-486

469

andMarx: Individualism Methodological on JonElster,GameTheory, SomeRemarks and OtherThings ROBERTPAUL WOLFF University of Massachusetts/Amherst Amherst, MA 01003 U.S.A.

In recentyears, philosopherstrainedin the techniquesand constrained by the style of what is known in the Anglo-Americanworld as 'analyticphilosophy'have in growingnumbersundertakento includewithin their methodologicalambit the theories and insights of KarlMarx. Twelve years ago, GeraldCohen startledthe philosophicalworld with a tightly reasoned analyticreconstructionand defense of one of Marx's most influential and controversialteachings, historical materialism.1 Seven years later, Cohen's friend and colleague, Jon Elster, produced what may fairlybe consideredthe definitiveanalyticphilosopher'sencounter with the thought of Marx.2 I find Elster'sbook to be fundamentally a failure, despite its many virtues. It seems to me almost entirely to miss what is important in Marx'sthought, frequentlyreducing it in the process to triviality, or, what is worse, to parody.3Now, stated thus baldly, my reactionmight fairlybe dismissed as a cheap shot, for Elsterfreely acknowledges the existence in Marxof depths and complexities which slip through his analyticalfilter. The problem, he thinks, is to find a way to translate Marx's rich, provocative, many-sided, but sometimes hopelessly insights, proposals,and metaphysicallytaintedtheories,asides, aperqus

1 G. A. Cohen, KarlMarx'sTheoryof History:A Defence(Princeton: Princeton University Press 1978) 2 Jon Elster, Making Sense of Marx (Cambridge: Cambridge University Press 1985) 3 David Schweickart has written a splendid review of the book which exposes both the inadequacies of some of Elster's scholarship and also the deeper political significance of Elster's anti-Marxian 'Marxism/ See Praxis International8 1988).

470

Robert Paul Wolff

rhetoricalflourishesinto somethingthat will withstandthe criticalscrutiny of a sympatheticmodern analyticphilosopher. Since I believe Elster has failed, it is incumbent upon me not only to gesture grandly in the directionin which Marx'ssuperiorwisdom seems to lie, but actually to state with some precision what Elster has missed, and how we might succeed in reclaimingit for our day. Beyond that, we must ask whether the models and forms of analysis which Elstertakes from the modern theory of rational choice are fundamentally unsuited to the task of making sense of Marx. Midway through the book, Elsterwrites the following words about his struggles with Marx'stheory of ideology. They could as well have served as a general summary of the deeper philosophical purpose of the entire book: In my struggle with Marx's writings on ideologies, I have been constantly exasperated by their elusive, rhetorical character. In order to pin them down, I have insisted on the methodological individualism set out [in the Introduction], with results that may appear incongruous to some readers. Yet I fail to see any satisfactory alternative. A frictionless search for the "function" of ideologies or the "structuralhomologies" between thought and reality has brought this part of Marxism into deserved disrepute. To rescue it - and I strongly believe there is something here to be rescued - a dose of relentless positivism seems to be called for. (239)

In the opening pages of the book, Elstersummarizesthe doctrineof methodologicalindividualismwhich he endorses. 'Bythis,' he says, 'I mean the doctrinethat all socialphenomena - theirstructureand their change - are in principleexplicablein ways that only involve individuals - theirproperties,their goals, theirbeliefs and theiractions. Methodologicalindividualismthus conceived is a form of reductionism'(5). Defendersof methodologicalindividualismcustomarilyground their position on the ontologicalclaim that only individuals are real, all else - corporations,institutions, states, societies - being in some way aggregates of individuals. Although Elster does invoke these considerations a page fartheron,4 they are not offered by him as the primary reason for his adoption of the individualist method. Rather,he says, the rationale stems from the fact that in scientific explanation, 'there is a need to reduce the time-span between explanans and explanandum - between cause and effect - as much as possible, in order to avoid spurious explanations (5).

4 Cf. where Elster writes: 'Methodological collectivism - as an end in itself - assumes that there are supra-individual entities that are prior to individuals in the explanatory order7 (6). Although he does not say so explicitly, it is clear from the context that he rejects the appeal to such entities.

MethodologicalIndividualismand Marx

471

I shareElster'scommitmentto methodologicalindividualism,but for the ontologicalreasons, which are, I think, more compelling and more constrainingthan those arisingfromconsiderationsof the requirements of scientific explanation. I am prepared to assert not merely that an explanationin terms of individualsis better, simpler, or in other ways more desirablethan an explanationin terms of such collective entities as states, classes, and institutions,but also that the inabilityto unpack such collectivistaccounts into their individualistcomponents of itself demonstrates that they can be no more than provisional sketches designed to guide us in promising directions. Nevertheless, we need to ask a question that Elster seems never to think to ask, and having asked it, to stay for an answer. Whydo serious, intelligent, clear- thinking social theorists like Marx - and like Emile Durkheimor KarlMannheim - appeal to a language and style of explanation that seems, at least upon first examination, to violate the canons of individualist methodology to which one might otherwise imagine them to be committed?Let us grant, at least provisionally, that these men really had something authentic and important in view, that they were not misled by a faulty discourse or inadequate grasp of the tools of analysis, but ratherwere in the grip of an insight that they were unwilling to relinquish merely out of methodological piety. Ratherthan speaking dismissively and with a regrettablecondescension of 'Marx'slack of intellectual discipline' (508), or of 'the omnipresent biasof wishfulthinkingin Marx'swork' (438), it would be much more useful to take it as a working hypothesis that Marxreally had his finger on something worth analysing, so that we might, by following along with him criticallybut generously, come upon understandings that otherwise might be denied us. I begin with what I take to be the pivotal passage in Elster'sbook. A bit more than midway through the text, in a section entitled 'The conditions for collective action,' the following appears: The motivationto engage in collective action involves, centrally,the structure of the gains and losses associatedwith it forthe individual... . The gains and losses associatedwith collectiveactionmust, for the present purposes, be measuredin terms of expected utility. Hence they depend both on the individual'sestimate of the likelihoodof success and failureand on the degree of risk aversion.5For

5 By alludingto degrees of risk aversion, Elsterimplicitlyinvokes the assumption that utility is cardinallymeasureable,with no apparentawareness of the enormously powerfulpremisesrequiredfor that assumption.Thereare even suggestions, as we shall see, of interpersonalutility comparisons.Here, as elsewhere, Elsteruses what I should call the rhetoricof game theory with no attention to its logic.

472

RobertPaul Wolff

the time being I assume that the utility derives from the materialgains and losses for the individual himself.... On these assumptions, then, the utility calculus of collective action is captured in three variables. The first is the gain from cooperation, defined as the difference between what accrues to the individual if all engage in the collective action and what accrues to him if none does. The second is the free-ridergain, that is the difference between what he gets if all but him engage in collective action and what he gets if everyone does so. Finally, there is the loss from unilateralism - the difference between what he gets if no one engages in collective action and what he gets (such as punishment or costs of engaging in useless individual action) if he is the only one or among the few to do so. Other things being equal, the probability of collective action increases with the first of these variables and decreases with the second and third. Frequently, however, they do not vary independently of one another.... In general, collective action will either be individuallyunstable(large free-rider gains), individuallyinaccessible (large losses from unilateralism) or both. Since nevertheless such action does occur, we must try to understand how these obstacles are overcome. (351-2)6

This passage perfectly captures the style and tone of Elster'sanalysis: superficiallycareful,precise,rigorous,apparentlyawareof the complexities of human motivation(in portions I have elided to save space, he recognizes the role of altruism,for example), a quantitativeformalism lurkingjust below the surface. Clearly,Elster'slanguage implies, if we insisted, he could put the whole thing into symbols, thereby removing the slightest vestige of subjectiveopinion from his analysis. And yet, the entire passage is utterly mad - a crackpotaccount that sounds as though it comes from Swift's account of the voyage of Lemuel Gulliverto Laputa,or from Anatol France'sPenguinIsland,or, worse still, from Robert Nozick's Anarchy,State,and Utopia. Think for a moment about what Elster is saying. Collective action, according to him, is individually unstable, individually inaccessible. As he says several pages later, in the midst of a discussion of the rationalityof collectiveaction, 'forcollectiveactionto take place so many conditions must be fulfilled that it is a wonder it can occur at all' (361). But the most casual survey of history and society shows us that collective action is the norm in human affairs. In every human group one can think of, collectiveactiondominates the waking hours - and even the sleep - of every one over the age of one and a half or two.

6 The notion of a 'difference' between two gains presupposes cardinal utility. Elster's formulation also makes sense only so long as there are no more than two strategies for each of two 'players.' Since, in general, there will be many strategies available to each of many players, the notions of gain from cooperation, freerider gain, and loss from unilateralism are incompletely defined. As we shall see, Elster is mesmerized by the elementary pictures of the Prisoner's Dilemma, and forgets to ask whether these little sums and differences correspond to anything in the real world.

MethodologicalIndividualismand Marx

473

A little reflectionwill remind us that all of the productive activities of human beings are collective in character,even those of the fabled RobinsonCrusoe.7All kinship interactions,sexual liaisons, all our activities of eating and warring, almost all religious activitiesand activities of artisticcreation, reproduction, and appreciation,are collective in character.Voting, strikes, military campaigns, riots, cocktail parties, family vacations - all of these, on Elster'sview, are so improbable that we can barelyunderstandhow they might, on rareoccasions, actuallyhappen. Clearly,there is something badly wrong with a theory of society that concludes that the norm is so abnormalthat it is almost never likely to occur! Where does Elster go wrong? Elster'sfirstproblemis that he never actuallydefines the phrase 'collective action,'despite the fact that the book is pretentiouslyquasi-formal, full of definitions, Game Theory jargon, allusions to payoff matricesand the like. Clearly,until we know with some precisionwhat he means by the term, we cannot even begin to evaluate his claimthat collectiveactionis unlikely, albeitactual,nor can we determinein what sense, if any, Marx'sexplanationsof collectiveexplanation,or anyone else's, have violated the principles of methodological individualism. The core of the argument, such as it is, can be found in Chapter6, 'Classes.' Elsterbegins by defining 'class.' After reviewing in a useful and interesting fashion some of the disputes that have grown up around the term, he offers the following definition: A classis a group of peoplewho by virtue of what they possess are compelledto engage in the same activities if they want to make the best use of their endowments.8To

say that classes are real,he explains a bit later, is to say that 'under certain conditions they tend to crystallize into collective actors,' and this latter phrase - crystallizinginto collective actors - is explained as meaning that they 'achieve class consciousness' (344). Thus far, not much light has been shed. In particular,we want to know what Elster understands by class consciousness. His ratherbizarreanswer is this: 'I define (positive) class consciousness as theability to overcomethe free-riderproblemin realizing class interests' (347).

This is, to put it mildly, not what Marxand other socialtheoristshave seemed to have in mind when they used the term'classconsciousness,'

7 Marx labelled the efforts by Vulgar Economists to read economic laws out of the imagined experiences of isolated producers /Robinsonades/ (Robinsonaden).See Marx-Engels Werke(Berlin: Dietz Verlag 1962 B. 23, 90. see also S.S. Prawer, Karl Marx and WorldLiterature(Oxford: The University Press 1976), 273 ff. 8 Elster, 331. Needless to say, this sounds more like a neo-classical than a Marxian definition, but so be it.

474

RobertPaul Wolff

but let us deal with Elsterin his own terms, for the moment, and see whether we can figure out what he is saying. There are two difficulties with the definitionhe offers of class consciousness: the first is that he does not tell us what the free-riderproblem is, and the second is that he does not explain what one would have thought would be for him the extremely problematicnotion of class interests. Considerfirstthe so-calledfree-riderproblem.Thereare actuallytwo free-riderproblems,not one. The firstis a problemfor those who want to get a group of people to act together in pursuit of some social, political, cultural,economic, religious, or other goal. The second is a theoreticalproblemof explanationfor rationalchoice theorists. Elster,like most rational choice theorists, confuses the two. The practicalfree-riderproblemis that sometimes, when we are trying to get a group of people to pursue a goal, it is hard to get everyone to pitch in and do his or her part, because some individuals may figure that it can't make any noticeabledifferenceif they slack off. Especially when the action involves considerableeffort, or cost, or danger, or when the connectionbetween the actionand the end isn'tvery clear, this sort of thinking may pose a serious threat to the success of the effort. My admittedlylimited experience suggests that relativelyrarely can the problembe tracedto deliberatelyselfish calculationsin which individuals literally figure out that their dominant strategy is nonparticipation.Notice that this is a practicalproblem which only rises to the level of concern when large numbers of people are slackers. A strike, an election, a riot, even a family picnic, can survive somelevel of free-riding,and experiencetends to teach us when that level is likely to be exceeded, and when it is not. To cite one actualexample from my own recent experience: when an organizationI run conducted a telephone campaign to get out a vote and raise money, we were told by the marketingfirm doing the calling that a 50%rate of pledge fulfillment was a reasonable expectation. Now, in fact, we experienced only a 40%fulfillmentrate, which createdsome financialproblemsfor us. But a 50%rate, which rationalchoice theorists do not even deign to discuss, would in real world terms have been quite satisfactory. The theoretical free-rider problem is this: if we make some very powerful, very restrictiveassumptions about the utility functions of a group of individuals and about the canons of choice to which they conformtheir decisions - assumptionswhich amount, roughly, to the premises that individuals make choices solely on the basis of expected benefitsto themselves, very narrowlyconstrued,thatthey imputeidenticalpreferencestructuresto all other individuals, that there is insufficientinformationor communicationor enforcementproceduresto affect individual choices, and that individuals choose so as to maximize expected benefits - then we can deduce that there are certain sorts of

MethodologicalIndividualismand Marx

475

actions, requiringthe active participationof large numbers of people, which will not occur. Strikes will be called, but no one will show up on the picket line. An election will be held, but no one will vote. The command will be given to charge the enemy emplacement at the top of the hill, but no one will move. A leader will cry, 'to the barricades/ but no one will budge. Now, the fact is that strikes, elections, infantry charges, and street rebellionsdooccur. So the theory of rationalchoice has a problem. Clearly, some of the premises of RationalChoice Theory are wrong. (Note: this is nota problemfor the strikeleaders, party bosses, Second Lieutenants,or revolutionaries.The group effortsthey are trying to promote are, ex hypothesi, occurring.The problem is for the theorists, who must confrontthe fact that their theories entail conclusions which are disconfirmed by the facts.) A little later, I shall address directly the question of just which premises of rationalchoice theory ought to be called into question. But first, we must try to decode the second phrase in Elster'sdefinition: class interests. What are class interests, accordingto Elster, and what does it mean to 'realize'class interests. An interestis a goal or end or aim or purpose that a purposive agent sets for itself (or, alternatively,that may, on some theory, be imputed to the agent despite the agent's unawareness of it. Elsterunderstands how importantthis addendumis, and has some intelligentthings to say about it. Since my disagreementswith him do not turn on this aspect of the subject, I shall ignore it here.) Any methodologicalindividualist - such as myself - will presumablyneed to define, or explain, class interests in terms of the interests of individual persons. It is thus extremely puzzling that Elster does not directly address himself to this task, apparentlyconsideringit self-evidentwhat a classinterestmightbe. Our best evidence of what Elsterhas in mind - and the real indication, I think, of the real use he wishes to make of rationalchoice theory - appears in the section of Chapter6 entitled 'The rationalityof collective action.' Here, at some length, is the passage: On first principles,one should seek for micro-foundationsfor collective action. To explain the collective action simply in terms of the benefits for the group is to beg all sorts of questions, and in particularthe question why collectiveaction fails to take place even when it would greatlybenefit the agents. The individuallevel explanationsshould be constructedaccordingto the followingheuristicprinciple: first assume that behavioris both rationaland self- interested;if this does not work, assume at least rationality;only if this is unsuccessfultoo should one assume that individual participationin collective action is irrational.... The basicproblemconfrontingany groupof people tryingto organizethemselves is that of the Prisoner'sDilemma. In its simplest form it is a strategicgame between any given individual and "everyoneelse." To each of these actors, two

476

RobertPaul Wolff

strategies are available: to engage in the collective action or to abstain. For any pair of strategies chosen by the actors, there is a well- defined payoff (in expected material welfare) to each of them. In the matrix below the first number in each cell represents "my" payoff and the second the pay-off to each of the individuals included in "everyone else." Table 6.1 Everyone else Abstain

Engage

I

Engage

I

b^b

I

e^f

I

Abstain

I

c,d

I

a, a

I

Here b-a represents the gain from cooperation.... Similarly c-b represents the freerider gain and a-e the loss from unilateralism. Clearly, whatever everyone else does, it is in my interest to abstain. If all others engage in collective action, I can get the free-rider benefit by abstaining and if everyone else abstains I can avoid the loss from unilateralism by abstaining too. Since the reasoning applies to each agent, in the place of "I," all will decide to abstain and no collective action will be forthcoming. In one sense the logic is compelling. If (i) the game is played only once, (ii) the actors are motivated solely by the payoff in the matrix and (iii) they behave rationally, collective action must fail. By contraposition, we might look into the possibilities for collective action if the interaction is repeated several times; if the payoffs that motivate the actors differ from the material reward structure; and if the behavior is less than fully rational. It turns out that under all these conditions, collective action does become possible. The three cases correspond to what is referred to earlier as rationality-cum-selfishness; rationality simpliciter; and irrationality. (359-60)

In light of these remarks,it would appear that by 'realizingclass interests' Elstermeans moving from the sub-optimalequilibriumof the Prisoner'sDilemma Game to the Pareto-preferredoutcome of mutual trust and cooperation. Implicitly(but onlyimplicitly),collective action is then action which achieves (or aims at? who is doing the aiming?) the Pareto-preferredoutcome. This simply won't do. Indeed, it won't do for so many differentreasons that it is a bit hard to know where to begin the critique.Forpurposes of organization, if no other, let me start with the most interior criticisms - those which accept Elster'sframeworkof analysis - and then proceed to call that framework itself into question. Let us begin, where Elster does, with the much-discussed, muchmisunderstood Prisoner'sDilemma. From a Game Theoreticpoint of

MethodologicalIndividualismand Marx

477

view, the little payoff matrixwhich he introducesinto the text is a complete mess. Here are some of the problems: 1. The matrixpurportsto represent payoffs in expected materialwelfare resulting from the four possible pairs of strategy choices. Elster forgets to tell us how the players rank these outcomes, so the matrix, as it stands, doesn't define a Prisoner'sDilemma game. Furthermore, since interpersonalcomparisons of utility are, I presume, not being posited, the use of the same letters (a and b) for payoffs to both players is extremelymisleading. To make the matrixrepresenta Prisoner's Dilemma, we must assume that the players have the following preference structure(I pass over the not insignificantfact that Elsterfails to distinguish between the rank ordering of quantities of materialwelfare and the rank ordering of preferences): for the player identified as %'c>b>a>e,and for the playeridentifiedas 'everyoneelse,' f >b>a>d. 2. A Prisoner'sDilemma is a two-person game with no communication. It is assumed that this is a situation of choice under uncertainty, which means that the outcomes are well-defined, the strategyoptions are well-defined, and the players'preferencesare well-defined,but the players have no way of estimating the probabilitythat other players will select particularstrategies. In the present case, this means that neitherplayercan make a reasoned estimateof the probabilitythat the other players will choose to coordinateon a policy of mutual engagement in collective action. But although this may very well model a laboratorysituationin which subjectsrecruitedfrom a universitycampus are run through little artificialgames, it completely fails to model the actual situation of a platoon, a union local, a family, or an electorate. Note: this is notto say that such real-lifegroups act 'irrationally.' Quite to the contrary - it would be wildly irrationalfor a group of voters, workers, or soldiers to ignore what they know about one another,what they rememberof theirpast interactions,and what they have communicated to one another. The Prisoner's Dilemma, mesmerizing though it may be, simply is not a model of group action. 3. A Prisoner'sDilemma is a game defined by a two-by-two matrix, which means that it is a game in which each playerhas only two strategies. It is actually very difficultfor those unfamiliarwith Game Theory to grasp just how reductively, absurdly, uninterestingly simple a game must be in order to offer only two strategies to each player. By way of example, consider the following silly little game, which I invented to make my point. There are two players, A and B, who start with a pile of four matchsticks. A move consists of removing either one or two matchsticksfrom the pile, and players move alternately,

478

RobertPaul Wolff

who goes first being decided by a coin toss. The player who removes the last remaining matchsticksloses. The longest the game can last is four moves - two for A and two for B. The shortest is two moves - one apiece. Not a very interesting game, certainly not as interesting as the game of getting eighty million people to vote, or the game of getting three thousand workers to strike, or even the game of getting eleven soldiers to charge a hill defended by a machine gun nest. And yet in this little game, A has twelve strategies, B has twelve strategies, and the payoff matrixis a twelve by twelve matrixwith one hundred forty-fourboxes (leaving to one side Nature's choice of heads or tails, which would require a thirddimensionto represent).The degree of simplificationand abstraction needed to construe a situation as capable of being modelled by a two-by-two matrixis such that there is almost certain to be no interesting connection between the model and any social, political, or economic reality.9 4. Elster puts on a fine show of formalist rigor with his equating of the quantity (b-a) to the gain from cooperation, and so forth. These quasi-quantitativeformulaehave any meaning at all onlyif we are talking about a two-person game in which each playerhas only two strategies. If, as must certainlybe the case in the real world, there are many players, each with many strategies, then the meaning of 'gain from cooperation'or loss from unilateralism'loses all precise meaning, and becomes a metaphor without a referrent. To see that this is so, consider a very slightly more realisticgame, called Strike.The game has three players: A, who is the strikeleader, and B and C, who are the followers. The game has four moves: A goes first, and either calls a strike, or doesn't. B who goes second, must go along with A if A doesn't call a strike, but may choose either to join or not join a strike if A calls one. C goes third, and has the same options as B, but with the difference that C knows what B has done. Finally, A goes again, and can either affirmor cancel a strike in light of what B and C have done, assuming that A has called a strike on the first move.

9 The game is much simpler if A automatically goes first. Then, A has 3 strategies, B has 4, and the matrix is 3 x 4. For those who are curious, here are the strategies available to the two players in this simpler game. A's strategies are (1) Take 2. If B takes 1, take 1. (2) Take 1. If B takes 1, take 1. If B takes 2, take 1. (3) Take 1. If B takes 1, take 2. If B takes 2, take 1. B's strategies are (1) If A takes 2, take 2. If A takes 1, take 1. If A then takes 1, take 1. (2) If A takes 2, take 2. If A takes 1, take 2. (3) If A takes 2, take 1. If A takes 1, take 1. If A then takes 1, take 1. (4) If A takes 2, take 1. If A takes 1, take 2.

MethodologicalIndividualismand Marx

479

In this little game, A turns out to have 17 strategies, B has 2, and C has 4. A payoff matrixwould thereforehave to be a 3-dimensional array, 17 x 2 x 4, with 136 boxes, in each of which would be entered a triadof numbers or letters representingA's, B's, and C's evaluations of the particularone of the nine possible outcomes arrived at by the intersectionof the three strategiescorrespondingto the row, column, and depth (or whatever) intersecting at that box. There is nothing in this game quite so simple as the gain from cooperation, the free-rider gain, or the loss from unilateralism,even assuming one could define the appropriatecardinal measures of utility. 5. Elstercompletelyconfuses his own formalismby describingthe second of each pair of payoff entries as representing the payoff 'to each of the individuals included in "everyone else/" But this simply won't do! If all the others are independent players, then this is an n-person game, not a two-person game. And if 'everyone else' really is a group acting as a group, with two strategies, then all those people have, ex hypothesi, solvedthe problemof actingcollectively,in which case what the odd person out does is of relatively little interest! To see how confused Elster'saccount actuallyis, consider this passage, appearingimmediatelyafterthe definitionof class consciousness as the abilityto overcomethe free-riderproblem.The problem,he says, is that 'the individual can reap a greater reward if he abstains from the action to get the benefits without the cost. This generates a conflict between the interest of the individual class member and that of the class as a whole. (347)10But what can the phrase 'that of the class as a whole' possibly mean for Elster?If the logic of free-ridingleads everyone to defect, then the problem is not a conflict between the interest of the individual and the interest of the class as a whole. On Elster'sown view, the problem is a sub-optimaloutcome for each individual. The conflictis between what the individual wants and what she gets. The notion of class interest does not enter. It is precisely here that Elster'sfailure to explain the notion of class interest leads him into confusion. It is clear that no aggregativefunction, such as average utility or total utility or a weighted average of utilities, is going to do the trick, even if we allow interpersonalcomparisons of cardinalutility. In fact, this problem helps us to see why

10 Elster then gives, as a supposed example of this, a passage from the Communist Manifesto, but in fact the passage quoted does not describe an example of the free rider problem at all. It alludes to the fact that competition for jobs in the labor market interferes with the 'organization of the proletarians into a class/ an entirely different matter.

480

RobertPaul Wolff

the Prisoner'sDilemmahas such an appeal for him. In that little game, there are only two players and no pre-play communication - hence, there are no aggregation problems and no possibilities of sidepayments, etc. In the n-person case, however, very sticky theoretical puzzles arise which powerfully resist plausible analysis by the models of Game Theory. We could continue to give instancesof Elster'smisuse, or lack of understanding, of the terminology and formalismof Game Theory11but it is more useful to try to locate the source of the inadequacy of his methodology. The real problem, I suggest, is an incorrectnotion of the self which is engaged in action, collective or otherwise, and a consequent inabilityto understand how individuals conceive of their situation, or formulate the goals of their action. Elster is quite right that we should, on principle, seek microfoundations, even though, as he points out, one must avoid 'premature reductionism'because 'collective action may simply be too complex for individual-levelexplanationsto be feasibleat the currentstage' (359). He is also correct, in my judgment, in asserting that we ought to begin by assumingthatbehavioris rational,althoughnot in the sense of being calculativelymaximizingbehavior.Rather,we should assume that behavioris rationalin the sense of being purposive, goal-oriented, guided by considerations of instrumentality - that the individuals whose behaviorwe wish to understandcould, at least in principle,give a coherent account of why they are acting as they are, by reference to what they seek to achieve and how they expect what they are doing to advance their goals. The problemstartswith Elster'sinclusion of the assumption that behavior is 'self- interested.'To see what is wrong with this assumption, let us returnto the Prisoner'sDilemma.Formallyspeaking, a Prisoner's

11 See, for example, the misuse of the term 'constant-sum game' on 373, and the completely garbled term Variable-sum game/ which has no meaning in the formal development of Game Theory. A two-person strictly competitive game is correctly describable as 'zero-sum' or 'constant-sum' under certain extremely powerful assumptions about the preference structures of the players - who, incidentally, can be classes of individuals only if one can give meaning to the notion of the preference structure of a class! Games which are not constant-sum can only be described as not constant-sum. The concept of a sum of payoffs is undefined for such games (because such a sum would involve interpersonal comparisons of utility). Hence, they cannot be said to be variable sum. Since it is precisely the notion of class interests rather than individual interests which Elster is trying to elucidate, it is especially misleading to throw these terms around with no awareness that their use begs precisely the questions about the nature of collective interests which are at issue. This is the way in which the unrigorous use of formalism conceals rather than dispels confusion.

MethodologicalIndividualismand Marx

481

Dilemma is a two-person game, with two strategies for each player, in which the playerspreferthe outcomesin the patternindicatedabove. Any two person game, with two strategiesper player, with preferences for payoffsconformingto this patternis a Prisoner'sDilemma.Butthere is a little story that gives this formalstructureits name, and that little story contains more informationthan ends up being encoded in the payoff matrix.It is that extrainformationthat is the source of the difficulty. The story, as everyone knows, concerns a pair of criminalswho are nabbed in a robbery, held separately in jail cells, and presented by the DistrictAttorney with a set of threatsand promises concerning the jail sentences they will receive if each of them does or does not turn state's evidence. The outcomes resulting from the criminals'various strategy choices are expressed in the number of years they may receive as sentences. The unspoken assumptionwhich underliesthe game is that each criminal ranks outcomes solely accordingto the length of his sentence (or hers, but the little story is always told about two men), preferringa shorter to a longer sentence. It is thus assumed that neither criminal is willing to serve as little as another minute in jail even in order to keep his buddy from going to the gas chamber. It is these assumptions that allow us to translate the story into a payoff matrix.12 Butrationalitydoes not requirethat individualsrankoutcomes in this way. Indeed, even se//-interest,broadly enough construed, does not imply the utility functions assumed to be operative in the Prisoner's dilemma.To see how things mightbe different,consideranothergame, which consists of two violinists playing the BachDouble Concerto.Let us suppose thateachviolinistcan choose between playingas fast as possible, or playing a tempo.Therearefour possible outcomes: if both play as fast as they can, the resultis a musicalfiasco, but a personalstandoff. Neither is humiliatedby having been shown up as incapableof presto playing. If both play a tempo,the result is beautifulmusic. If one plays fast and the other plays a tempo,the result is personal triumph for the first and humiliationfor the second, and, of course, musical disaster.

12 Formally speaking, all of this amounts to stipulating that each player's utility function is inversely monotonically related to sentence length, or, alternatively, that each player's utility function is a lexicographic function which first minimizes that player's sentence, and only then responds to other variables. Without assumptions like these, we can construct an outcome matrix which specifies what each player gets for each pair of strategy choices, but we have no way of translating that outcome matrix into a payoff matrix. The point of the phrase 'as little as another minute' is that with only ordinal rankings (and no interpersonal comparisons of utility), one cannot say anything about how much one is giving up in relation to how much one is inflicting on the other player.

482

RobertPaul Wolff

Now imagine two pairs of players playing this game. The first consists of David and IgorOistrakh,who in factproduceda transcendently beautiful recording of the composition, and the second consists of BeverlySumac and myself at Mrs. Zacharias'sannual parlorrecitalfor her violin pupils in the late spring of 1946. Beverlyand I, I will simply report, treated the event as a Prisoner's Dilemma (in more senses than one!), with the result that we raced as fast as we could to the end of the piece. Since she was on her way to a career as a concert performer, and I, to put it mildly, was not, she won the race. The result was not music. The Oistrakhs, on the other hand, had a different preference structure. They were engaged in a collective activity - the making of beautiful music. For them, the joint playing of the concertoa tempowas the most preferred outcome, a madcappresto performance,we may imagine, came next, and the two other outcomes were treated as indifferentlyworst. The result was a coordination game in which the purely game theoretic aspects of the coordinationwere triviallyeasy, inasmuchas each could only lose by defecting from the strategy of playing a tempo(I leave to one side the non-game theoretic aspects of the effort to play the Bach Double, which were also easy for them, but would be distressingly difficult for me). Note that there is no question here of altruism, or self-denial. Indeed, althoughIgorwas David'sson, they could just as well have been mortalenemies, so far as the problemof coordinationwas concerned. Each is engaged in goal-oriented, purposive, self-interestedaction i.e., each is acting rationally. But since the goal that each pursues is the mutual and collective making of beautiful music, they coordinate rather than frustrate one another. The same point can be made with regard to the free-riderproblem. It must puzzle Elster how a large symphony orchestracan ever play the BrahmsSecond. After all, we can imagine him reasoning, no one but a Toscanini - and certainlynot Seije Ozawa - will notice if a single second violinist puts soap on his bow and only pretends to play. But then, by parity of reasoning, the entire second violin section will soap up, and the orchestrawill fall flat. No doubt some orchestralmusicians, long in the tooth and cynical besides, might reason this way. But most orchestraplayers have, as their first preference, to get paid for a first-class performance in which they participate.Professional violinists do not begin to exhibit negative marginalutility for playing the violin until long past the limits of a Brahmssymphony (I leave to one side the question of a Mahler symphony). It is entirelypossible - indeed, it is, I suggest, usually actual - that men and women will have, in this sense of the term, collective goals in the pursuit of which they engage in collective action. Nothing in

MethodologicalIndividualismand Marx

483

the having of such goals requiresus to posit any entitiessave individual persons, nor is the pursuit of such goals in any sense irrational.Indeed, the pursuit of such goals is not even altruistic or non-selfinterested. David Oistrakhdoes not play a tempoout of a selfless putting of his son's interests ahead of his own. He plays a tempobecause he aims at a goal - the collective creationof beautiful music - which cannot be reached in any other way. But, Elsterwill reply, romanticsentimentalityto one side, a worker does not engage in strikes because he has adopted it as his goal to raise everyone's wages in a just and equal manner through joint action. If he did, there would beno free-ridergain. He engages in strikes in order to raise his wages. The raising of everyone else's wages may be a necessary means to his end - one which he will therefore support. But on the assumption that there are costs associated with committing oneself to a strike, he will ever be on the lookout for a way of obtainingthe benefits - higher wages for himself - without incurring the costs. Now, the plainfactof the matteris thatthis objectionis, as a universal generalizationaboutthe behaviorof workers - or other groups of people - just plain false. Most human behavior, it seems to me - the mean-spirited, ugly, cruel, unjust behavior included - is motivated by what may be called the pursuit of social, or collective ends. The reason for this is that human personalityis formed by the internalization of social norms and roles, and by the identification of self as a memberof familial,religious, geographic, political, military,social, or cultural groups, so that most people, most of the time, understand themselves and their situations in terms of the groups in which they are most securely imbedded. The conception of self on which rational choice theory bases its assumptions about individual motivation and choice are not only historicallyand culturallyquite specific. Even in those cultures, and at those times, when individuals learn - culturally - to exhibitwhat Elstercallsrationalbehavior,they exhibitit in very severely constrained ways and with regard to very narrowly limited ranges of options. The conscious regulation of conduct by calculationsof self-interest is so rarein human history that the greatest sociologists of the classical period - Marx,Weber, Sombart, and the rest - devoted endless efforts to explaining its appearanceat a particularhistoricalmoment in the evolution of western Europeansociety. So unusual is such conduct even in capitalistsocieties that when we encounter an individual who allows rationalcalculationto regulate more than a narrowly circumscribedsphere of economicdecisions, we arelikelyto considerhim or her seriously pathologicallyderanged. Paul Goodman capturedthe crackpotqualityof calculativeinstrumentalrationalityrun amok in the

484

RobertPaul Wolff

characterof the mercantilecapitalistEliphazin his wonderfully satirical burlesque of capitalism and modern educational theories, Empire City. What makes Eliphaz so wacky and non-human is precisely that he resists the natural tendency to engage in collective action. Oddly enough, Elster knows that something is wrong with his attempts to explain collective action by appeals to iterated Prisoner's Dilemmagames and such like ephemera.Afteractuallycanvassingthe possibility that workers'utility functions are influenced by 'externalities' (economists' jargon for whatever doesn't fit the Procrusteanbed of economic reasoning), he writes: Workers no less than capitalists might engage in collective action because they find it selfishly rational. I find it hard to reconcile this idea with the extensive literature on working-class culture, but on the other hand the elusiveness and subtlety of these problems of individual motivation should make us wary of dismissing it out of hand. (363)

It is hard to know what to say to an author who finds the most commonplacesortsof human motivation'elusiveand subtle,'and yet thinks that something as esoteric as egoistic maximizationof expected utility is so transparentthat it can simply be taken, unexplained, as a datum of explanation. Let me offer an analysis of class consciousness and collective action as an alternativeto Elster's.This analysis will necessarilybe brief, but it is taken from a somewhat longer essay published more than twenty years ago, and those who are interestedcan consult the fullerversion.13 If we adopt Ralph BartonPerry'suseful definition of a value as any objectof any interest,then we can say that RationalChoice Theory is designed to analyse the choices of individuals who pursue egoistic values,or, equivalently, who aim at objects, events, or states of affairs in which they take an egoisticinterest.An egoistic interest is an interest which relates solely to the subjective state of the individual him- or herself. The goods and services flowing from economic activity, for example, are assumed by economists to be enjoyable by the solitary consumer, and to be valued for that reason. Needless to say, the goods cannotbe producedby the individualindependently,but what the consumer values is the consumption of the goods and services, to which end the economic system is merely a highly efficient means. Egoism as a theory of the nature of value is not to be confused with the assumptionthatindividualsact selfishly.One can hold thatallvalue

13 See Chapter Five, 'Community/ in Robert P. Wolff, The Povertyof Liberalism(Boston: Beacon Press 1968).

MethodologicalIndividualismand Marx

485

is egoistic and yet prescribe,or claim that men and women do in fact practice, altruism. The simple altruistbelieves that all value is egoistic. He or she simply seeks to maximize someone else's value. There is, however, another class of values, or of valued states and experiences, which we may call socialvalues. These are states of affairs whose realizationdepends essentially (not merely instrumentally) upon a reciprocalrelation between another's experience and my own. The example most familiarfrom the literatureof politicaltheory is the master/slaverelationshipdescribedby Hegel. Those who wish to be masters - as opposed to those who merely desire the private satisfactionsthat were once provided by servants, and now more and more are provided by labor-savingmechanicaldevices - those, that is to say, who want the experience of mastery, require sentient, purposeful, servantswho are subservient,and who conceiveof themselves as subservient. It is impossible to describe mastery adequately without making referenceto the servant's awareness of his subservience. Thus, if one sought to be a master,and by a peculiaraccidentsucceeded only in bending to one's will a flagellant who, for his own religious ends, was using the relationshipas a means to achieving a saving humiliation, one would entirely fail to achieve one's goal. Somewhat more to the point, those who pursue democracyfor its own sake seek to bring into existence a state of affairsin which free and equal men and women engage in rationaldiscourse for the purpose of choosing, and then realizing,jointlyarrivedat ends. Forthem, the process of free deliberationis itself valuable, over and above the ends which may thereby be attained. I suggest that collectiveconsciousness is that state of affairsin which all or most of the members of a group take an interest in, or aim at, the same social value, and know that the others are doing so. Collective action is then the cooperative action of a group of people in pursuit of the actualizationof some social value. Class consciousness, in particular,is the pursuit, by all or most of the members of an economic class (however defined) of the state of affairsin which the members of the group achieve economic well-being and politicalpower, andare mutually awareof having done so through their cooperativeand collectiveef-

forts.That mutual awareness is a part of what is aimed at, and hence the value that each seeks to actualize is inseparable from it. Thus understood, collective action is neither mysterious nor methodologically suspect. What needs explaining - what Marx undertookto explain in the context of mid-nineteenthcentury European politics and economy - is how, why, and under what constraints a group of individualscome to take an interestin particularsocialvalues. The Prisoner'sDilemmaand the Free-RiderProblemare inappropriate analytictools for understandingclass consciousness because both

486

RobertPaul Wolff

of them assume that in their preference structures, individuals pursue only egositic values. That assumption, which underlies all classical and neo-classical economic theory, is in fact so restrictive and incompatiblewith the common place realityof human experiencethat it provides no firmbasis at all for an explanationof what we commonly understand as collective action. ReceivedApril, 1990