- Email: [email protected]
Sentimental fools: a critique of Amartya Sen’s notion of commitment
Journal of Economic Behavior & Organization Vol. 40 (1999) 373–386
Sentimental fools: a critique of Amartya Sen’s notion of commitment Elias L. Khalil ∗ 1338 F. Hyde Park Blvd., #3, Chicago, IL 60615, USA Received 25 October 1997; received in revised form 8 February 1999; accepted 11 March 1999
Abstract Commitment is problematic because one sometimes pursues it against one’s interest. To solve it, the paper proposes a distinction between ‘non-binding’ and ‘binding’ commitments. Non-binding commitment is about ambition, such as becoming a great chef, which bolsters welfare in the pecuniary sense as well as self-respect. In contrast, ‘binding commitment’ is about honesty. While it diminishes welfare, it augments self-integrity. The neoclassical view reduces both commitments to interest, while the multiple-self approach separates both commitments from interest. The separation permits the confusion of sentimental fools, who enter commitments without regard to interest, with rational sentimentalists, who take interest into consideration. ©1999 Elsevier Science B.V. All rights reserved. JEL classification: D0 Keywords: Substantive interest; Non-binding commitment; Binding commitment; Sentimental fools; Rational fools; Rational sentimentalists; Guilt; Disgrace (shame); Embarrassment; Substantive preference; Symbolic preference (self-respect and self-integrity); Lethargic will; Crooked will
1. Introduction Given that commitment appears to negate interest, some theorists, such as Amartya Sen, propose that agents have conflicting selves. Defenders of the standard approach, on the other hand, deny the existence of the problem; they simply regard commitment as an ordinary preference stemming from a unitary self. While this paper recognizes the phenomenon of commitment as a problem, it tries to solve it without appealing to the multiple-self view. ∗ Tel.: +1-773-548-7638; fax: +1-773-702-8490 E-mail address: [email protected] (E.L. Khalil)
0167-2681/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 2 6 8 1 ( 9 9 ) 0 0 0 6 0 - 8
374
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Fig. 1. The three preferences.
It does so by distinguishing two kinds of commitment and how each relates differently to interest–a necessary distinction to discern admirable acts from sentimental foolishness. Let us start with three kinds of preferences, which may overlap, facing the self (S): Situation 1: S decides to enter a training program in the culinary art of French cuisine because the going wage rate for chefs, given the cost of training and the responsibility of the job, is more attractive than other options. Situation 2: In addition to Situation 1, S has been impressed by the creative art involved in French cooking. He takes on a commitment to try to become the best chef in the city. Situation 3: In addition to Situation 2, S takes on the commitment to be truthful. In specific, he prepares, in light of his advertisement, dishes made only from fresh ingredients, even when the pecuniary benefit from a cheating strategy is unambiguously greater than expected pecuniary cost resulting possibly from a tarnished market reputation. Situation 1 involves what I call ‘substantive interest’, pecuniary calculation. Situation 2 consists of ‘non-binding commitment’ because the promise cannot be obligatory given that the agent is uncertain whether he can fulfill it. Situation 3 entails ‘binding commitment’ because the agent is certain that he can meet his obligation. The agent fulfills his binding commitment to enjoy self-integrity. If he fulfills it exclusively out of substantive calculation concerning penalty, it would fail to generate the sense of self-integrity. While Situation 2 advances substantive preference, Situation 3 by definition reduces it. Nonetheless, both kinds of commitment satisfy a sense of selfhood, dubbed here ‘symbolic preference’ as opposed to ‘substantive preference’. As summed up in Fig. 1, while the negation of one’s substantive interest would be irrational, the negation of either kind of commitment is usually described as the result of weak will. There are two kinds of weak will (‘lethargic’ and ‘crooked’), and two kinds of symbolic preference (‘self-respect’ and ‘self-integrity’), following the two kinds of commitment. The paper commences with some motivation (Section 1). It distinguishes binding from non-binding commitments (Section 2) and preference from choice (Section 3). Many authors confuse non-binding with binding commitment, which lead them to separate commitment per se from substantive interest (Section 4). Such a separation entails that the agent is constituted of multiple selves (Section 5). Such a view, in turn, makes it impossible to distinguish sentimental fools from rational sentimentalists (Section 6).
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
375
2. Motivation While the three situations are about self-concern, one can equally illustrate them by other-concerns such as altruism, public spiritedness, or patriotism. The conjectured distinction among the three preferences has two payoffs, practical and theoretical. Regarding the practical payoff, the debate about income redistribution can be clarified on the ground of whether it is prompted by altruism, which pertains to substantive interest, or is instigated by fairness, which involves binding commitment. If a redistribution policy is prompted by altruism, one has to weigh the efficiency advantage versus side-effect disadvantages such as the weakening of incentives. But if the policy is prompted by fairness, on the basis of some social contract entered under the veil of ignorance (Rawls, 1971), one would conceive differently the so-called weakening of incentives: It would be in one’s right, as specified by the social contract, not to work below a guaranteed social wage. If there is no difference between commitment and interest, the judicial system’s main goal would be the deterrence of crime rather than the vindication of justice. Therefore, it would be cheaper to punish rich criminals with high fines rather than with jail sentences–as long as both methods are equivalent in deterring crime. However, people usually abhor the idea of fining rich criminals rather than putting them in jail. The abhorrence is inexplicable if one views commitment as a substantive preference. Moreover, the distinction between the two kinds of commitment will make it clear, as shown below, why the negation of non-binding commitment is only a cause for embarrassment, while the negation of binding commitment is usually seen as shameful. In fact, the distinction would provide a ground for why laws commanding charity, as distinct from upholding justice, can be counterproductive. In addition, if one views internal organs, such as kidneys, to be ‘gifts of life’, one would regard the legalization of their trade as repugnant, no different from prostitution (Radin, 1996; cf. Arrow, 1997). One would see the legalization of the market for organs as a violation of binding commitments which the agent, although implicitly, makes with one’s sense of identity. Some experts argue that legalization of the organs market would actually decrease the total organs available similarly to how the legalization of blood donation in the US has hindered the supply of blood (Stewart, 1992). While substantive preference rises as a result of the legalization, it may not rise enough to offset the weakening of symbolic preference which acts as a motive in institutions where the marketing of organs is prohibited. Regarding the theoretical payoff, the focus here, the proposed three-way distinction promises to supersede the shortcomings of both the unitary-self and multiple-self approaches. 1 The standard, unitary-self view does not recognize the interest/commitment distinction. When commitments are not the outcome of strategic repetitive games, the standard approach treats them as preferences reducible to substantive preference function which 1 I use the term ‘multiple-self’ to express inconsistent preferences, where the agent is assumed to contain a ‘moral self’ as opposed to ‘material self.’ In this sense, my usage differs from Richard Posner’s (Posner, 1995, ch. 4). Posner takes the term to denote conflicting needs, where agents are seen as trying to accommodate pressing needs arising from work, family, church, and so on (also see Kuran, 1995, pp. 43–44]. In the sense of inconsistent preferences, the multiple-self view is advanced by Amartya Sen (1982), John Harsanyi (1955), Thomas Schelling (1984), Amitai Etzioni (1986), Gregory Kavka (1991), and Jon Elster (Elster, 1986; cf. Hausman and McPherson, 1993). It is beyond the scope here to contrast fully the multiple- and unitary-self approaches (Khalil, 1995, 1996, 1997b, 1998).
376
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
is the object of optimization. On the other hand, the multiple-self approach recognizes the interest/commitment difference. But it treats non-binding commitment as not different from binding commitment. The treatment entails that commitment stems from a self separate from the one which expresses substantive interest. Such a view, as detailed below with respect to Sen’s, makes it impossible to distinguish sentimental fools from rational sentimentalists. The paper evaluates the advantages of the proposed three-way distinction vis-à-vis Sen’s work. In his 1977 celebrated paper, Sen [(Sen, 1977) (reprinted 1982, ch. 4] argues that the reduction of commitment to interest implies that agents are ‘rational fools’ because they are ready, in response to change of relative price or income, to cheat without the feeling of shame (disgrace) and to quit projects at the first appearance of difficulty without the feeling of embarrassment. It is true that casual observations of normal agents who occasionally cheat and, as soon as the project becomes harder, quit indicate that they experience shame and embarrassment--an experience alien to the rational fools inhabiting the unitary-self world. However, Sen’s separation of commitment from interest invites from the rear window another kind of fools, what is coined here ‘sentimental fools’. Sentimental fools are ready to incur costs unjustified by expected benefits, such as jumping off a ship in the middle of the ocean to retrieve a toy for a crying child. Sen is correct that the collapsing of commitment with interest engenders a world of rational fools, but his radical separation of interest and commitment engenders a world of sentimental fools. To avoid rational fools à la unitary-self view and sentimental fools à la multiple-self view, we should neither collapse interest and commitment nor separate them radically. It is possible to achieve this task with the help of the proposed distinction between non-binding and binding commitments.
3. The commitment dichotomy The difference between the two kinds of commitment is not that binding commitment is coerced, while the other is optional. Both are voluntary. No one coerces S in Situation 3 to advertize his fresh ingredients. Moreover, the difference between the two is not that binding commitment is a promise to a second party, while the other is a personal promise. Binding commitment can be made with the self, as when one refuses to accept payment for blood donation or sexual pleasure. Also, the issue is not that binding commitment is a non-binding one made in the past. All non-binding commitments move into the past once they are made. Rather, the main difference lies with regard to ability. In Situation 1, S is maximizing substantive preference which is assumed to be within one’s ability. In Situation 2, S is also pursuing substantive preference, but not in the strict optimization sense. The goal is rather expressive of S’s self-image, an image to which S makes a non-binding commitment. The commitment is non-binding because S is uncertain whether S has the ability to sustain the necessary effort to carry it through. S may eventually have to give up the commitment if it impairs sufficiently his substantive interest. However, insofar as S acts to attain such a goal, it affords S symbolic preference which goes beyond substantive preference. In contrast, in Situation 3, S is obliged by a binding commitment–pursued even in single-shot games–not to cheat or disown one’s identity, which can be extended to be loyal to one’s family, country,
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
377
tribe, and religion. The commitment is binding because S is certain that S has the ability to sustain the necessary effort to carry it through. In other words, while the agent’s ability is not in question in binding commitment, it is so in non-binding commitment (Khalil, 1997a). In binding commitment, the agent promises to be honest, not to sell or buy sexual pleasure (prostitution), not to accept money for donated blood, not to accept bribes, or to sell his country’s secrets to the enemy when it is clear that such a promise is within his ability. Given that ability is not in doubt, the reason behind the commitment, therefore, is to counter the opportunistic gain which would arise from cheating. In contrast, in non-binding commitment, the agent promises to pursue a career or advance a public cause despite anticipated arduous hurdles. The promise is needed in order to sustain one’s effort given that the ability of the actor with regard to the challenge is uncertain. This situation is similar to the commitment to climb a mountain, whose height and terrain are quite known to the actor, but would be an uncertain challenge for him because he has not tried it before. If the actor quits after trying hard enough, the actor would not be labeled (by himself or others) a ‘quitter’, i.e., someone who lacks fortitude. In contrast, the actor would be labeled ‘corrupt’ if he, as long as he is not in a state of destitute, accepts a bribe irrespective of its size. While the commitment to climb the mountain is contingent on an uncertain ability and hence is non-binding, the commitment to rebuff bribes is not contingent on ability and hence is binding. Further, each kind of commitment relates differently to substantive interest. While Situation 1 and Situation 2 augment substantive preference, Situation 2 additionally attains symbolic preference, in specific ‘self-respect’, for exhibiting tenacity. In contrast, Situation 3 originates symbolic preference, in specific ‘self-integrity’, for acting honestly, but only at the expense of lowering substantive preference as clearly in the case of single-shot games. 2 One may appeal to unreflected habit (Mueller, 1986) in explaining why one would act honestly in single-shot games. However, such an appeal is an option of last resort because it invokes ad hoc reasoning. A better explanation is that the agent in Situation 3 is optimizing a function which includes, besides the familiar substantive preference, self-integrity. While self-integrity arises from honesty which may lower substantive interest, self-respect stems from pursuing an ambition which enhances substantive interest. Another way to distinguish between the two commitments is to discern two kinds of weak will or temptations. When one gives up the pursuit of a non-binding commitment at the first sign of difficulty, as in Situation 2, his will can be characterized as ‘lethargic’. This differs from violating binding commitment, as cheating customers in Situation 3, where the will can be depicted as ‘crooked’. While lethargic will engenders the feeling of ‘embarrassment’, crooked will gives rise to the feeling of ‘shame’ for the simple reason that lethargic will invites complacency while crooked will is about duplicity. One can, with respect to binding commitment, be characterized by a crooked will while, 2 The proposed term ‘symbolic preference’ resembles Timur Kuran’s (Kuran, 1995) ‘expressive utility’ (see Khalil, 1999). There is one difference, however. The term ‘symbolic’ is merely an umbrella category, while Kuran’s ‘expressive’ is a concept which supposedly denotes a single preference function. To wit, Kuran, similar to many other writers on the subject, does not distinguish between the two radically different kinds of commitment and the consequently different kinds of symbolic preference. Aside from lexicon, symbolic preference is a category rather than a concept which denotes a phenomenon.
378
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
with regard to non-binding commitment, be characterized by a non-lethargic, tenacious will. 3 Anxiety-driven actions which may indicate neurosis (see Lewis, 1971) are motivated by guilt. Such actions may fail to highlight the proposed difference between the two kinds of weak will. The term ‘guilt’, as used below, exposes situations where action is motivated by anxiety rather than by preference. The term denotes suboptimal preference which one may choose out of lack of self-assertion, boldness, or self-confidence. As Charles Taylor (1995, pp. 142) argues, turning away from the crushing burden of medieval Catholic notion of ‘sin’ or, in contemporary vernacular, ‘guilt’, amounts to liberation: “To break with a good to which one cannot really subscribe is a liberation in anyone’s language.” There are other ways to establish the commitment dichotomy. At least concerning the three ways mentioned so far, they do not provide a clear-cut blueprint on how to classify unambiguously actual preferences. Many acts, such as volunteering resources to a homeless shelter, can be composite ones, blending the motives of altruism (i.e. substantive interest) with ambition (i.e. non-binding commitment to take pride in one’s neighborhood) and justice (i.e. binding commitment to the unlucky ones in society). At least theoretically, however, it is possible to discern the three preferences and two kinds of commitment they entail.
4. The divergence of preference and choice Among others, Becker (1996), Kuran (1995), and Sunstein (1997) propose other reasons, beside guilt-driven anxiety, behind the preference/choice divergence. To note, the set of concepts offered by each author is geared to solve a particular problem, rather than to solve the preference/choice divergence per se. For example, Becker, with other co-authors, tries to account for addiction, where agents choose bundles which are contrary to their preferences. He reasons that the preference/choice divergence is only apparent: Addiction, similar to habits, is the outcome of rational choice when consumption is intertemporally nonseparable. His account implies that observed preference/choice divergence is really the divergence of choice from, what I dub here, ‘fanciful preference’. In the case of addiction, the addict may wish to quit, but his preference might be fanciful when, given the hardship of quitting, it is beyond his budget constraint. The preference for a drug-free life style might be as fanciful as the dream of a school teacher with limited income to own a mansion in the French Riviera. Becker’s theory should not be misunderstood as stating that it is impossible to quit addiction or to live in a mansion with servants. Rather, to attain such objectives, the addict or the school teacher must expend enormous cost to cultivate appropriate new habits, such as socializing with non-addicts or augmenting additional human capital. Otherwise, to close the gap between choice and fanciful preference, the agent must adjust the fanciful preference as he learns from information. Further, Becker’s theory should not be misunderstood as failing to recognize the possibility of pathology, the persistence of the gap between choice and fanciful preference. But such persistence is the result of the misfunctioning of information processing. So, the 3
The analyses of Davidson (1980) and Jon Elster (1983) mostly focus on crooked will.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
379
persistence of the gap is reasoned to be the outcome of special characteristic of information, not expressive of a gap between choice and feasible preference. Becker’s reasoning shows that if one wants to assert the preference/choice divergence thesis, one must first treat preference as feasible. This seems to be the case with Frankfurt’s (Frankfurt, 1971) notion of ‘free will’ with regard to ‘second order preference’ or what came to be called ‘metapreference’ (also see Jeffrey, 1974; McPherson, 1980; Hahn, 1982). One may prefer choosing drugs over education, while one simultaneously prefers the preference for education over drugs at the meta level. For such a metapreference to be non-trivial, George (1993) explicitly distinguishes metapreference from ‘regret’ which is associated with particular choices. In such choices, regret express situations where alternative options are fanciful, i.e., beyond one’s budget constraint. On the other hand, Kuran studies the preference/choice divergence with a focus on the dynamics between private preferences and hostile public opinion. He argues that agents may not act on their private preferences if they do not jibe with public preferences. Kuran’s main objective is not to solve the preference/choice divergence in general. Rather, he tries to account for abrupt changes in public preferences. To Kuran, agents may ‘live a lie’ in public, falsify their private preferences to secure the approval of others. Such ‘preference falsification’ is justified i.e., not judged as indicative of insincerity or dishonesty only when the disapproval of the public involves severe persecution. For Kuran’s purpose, the preference/choice divergence is relevant only insofar as it sheds light on the divergence of private preference from public choice. Thus, this particular divergence stems from historical accidents of coercive social-political environments. In the absence of such an environment, and especially when public opinion values maverick and eccentric private preferences, there is no need for public choice to diverge from private preference. Nonetheless, Kuran’s thesis is useful here. It clarifies one issue. If one wants to assert the preference/choice divergence thesis, one must be clear that the issue is about truthful choice where the approval of others is irrelevant. Further, Sunstein investigates the preference/choice divergence with a focus on how societal corrupting norms may push choice away from true preference. He maintains that preferences are fuzzy and, hence, susceptible to the manipulation of unenlightened societal norms such as the pursuit of vanity. This entails that there is a need for state intervention in order to align choices with true preferences. Sunstein’s thesis about the deviation of choice from true preference does not fully coincide with the preference/choice deviation discussed here. For one thing, there would be no divergence if the public is fully enlightened even to the satisfaction of the benevolent state. Nonetheless, Sunstein’s discussion clarifies one issue. If one wants to assert the preference/choice divergence thesis, one must be clear that the issue is about the divergence of choice from enlightened preference.
5. Lumping non-binding with binding commitments Sen (1982, p. 1; 1985, 1993, 1995) proposes the preference/choice divergence from a different angle. He argues, contrary to the axioms of the theory of revealed preference, one should not assume a priori that behavior is internally consistent. Empirical study should rather determine whether, in a particular case, choice matches preference. While the
380
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
paper does not quibble with this thesis, Sen extends the thesis to imply that the agent has multiple selves. This implication is unwarranted. To show this, we need to unpack Sen’s confusing categories. The confusion arises from a terminological twist which equates the preference/choice divergence with the interest/commitment distinction. He reaches such an identification by equating the terms ‘preference’ with ‘interest’, on one hand, and ‘choice’ with ‘commitment’, on the other. While the issue is not about lexicon, this paper employs ‘preference’ to denote both interest and commitment, and utilizes ‘choice’ to indicate action which may deviate from preference. When Sen equates preference with interest and choice with commitment, he can claim with such a terminological twist that the interest/commitment distinction is necessarily implied by the accepted preference/choice divergence. With this lexicon, Sen overlooks the commitment dichotomy when he lumps both kinds of commitment in one concept and, in turn, opposes it to interest as if each kind of commitment relates in the same way to interest. For instance, he explicitly opposes ‘commitment’ and ‘sympathy’. He illustrates it by distinguishing two separate motives behind the opposition to torture: one is out of principle and the other is out of sympathy in the sense of caring about the substantive interest of the other (Sen, 1982, pp. 91–92). He further illustrates the difference: The contrast between sympathy and commitment may be illustrated with the story of two boys who find two apples, one large, one small. Boy A tells boy B, ‘You choose’. B immediately picks the larger apple. A is upset and permits himself the remark that this was grossly unfair. ‘Why?’ asks B. ‘Which one would you have chosen, if you were to choose rather than me?’ ‘The smaller one, of course’, A replies. B is now triumphant: ‘Then what are you complaining about? That’s the one you’ve got!’ B certainly wins this round of argument, but in fact A would have lost nothing from B’s choice had his own hypothetical choice of the smaller apple been based on sympathy as opposed to commitment. A’s anger indicates that this was probably not the case (Sen, 1982, p. 93) also see (Sen, 1993, p. 501). 4 It is clear that Sen places ‘sympathy’ in the substantive interest (welfare) function, while he treats ‘commitment’ as a category which reduces that function. Sen’s interest/commitment opposition makes sense if the word ‘commitment’ strictly denotes binding commitment. However, Sen uses the term to also denote non-binding commitment such as the commitment ‘to exert’ oneself, which rather augments rather than reduces substantive interest (Sen, 1982, pp. 97–98). Sen uses the same term to discuss the binding rule of justice to abstain from the use of torture and to express striving towards challenging goals, as working harder for a better future. Sen must have treated non-binding commitment as binding when he argues in many places that commitment per se always lowers substantive interest (welfare) (e.g. Sen, 1982, p. 93). For Sen, ‘commitment does involve, in a very real sense, counterpreferential choice’ [Ibid.]. Such an interest/commitment opposition overlooks that non-binding commitment 4 Sen (1997, p. 747) repeats the same observation with regard to the tendency of an agent not to grab immediately his favorite chair in a party or snatch his favorite fruit from the host’s tray. Sen dismisses the view that such etiquettes are ways which allow an actor to sympathize with the intensity of preferences of others [Ibid., p. 747n]. He maintains that it arises either from the concern over reputation, social commitment to a moral imperative, or conventional rule following.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
381
Fig. 2. Sen’s meta-ranking.
buttresses interest. As discussed next, the opposition leads to a particular view of ranking of preferences, the multiple-self view. As shown later, such a view cannot afford the conceptual tools needed to distinguish sentimental fools from rational sentimentalists.
6. Meta-ranking of preferences and the multiple-self view Sen welcomes John Harsanyi’s (Harsanyi, 1955, 1986, 1988) distinction between ‘ethical’ and ‘subjective’ preferences. While the former expresses what the person prefers ‘on the basis of impersonal social considerations alone,’ the latter denotes what the person actually prefers ‘on the basis of his personal interests or on any other basis’ [Ibid., p. 315]. This sociological approach generally defines the ‘good’ from the social rather than the individual standpoint. For Sen (1982, p. 100), however, Harsanyi’s ‘dual structure is deficient.’ First, ‘actual’ behavior may express ethical consideration and, second, ‘a preference ordering can be more ethical than another but less so than a third.’ Sen proposes ‘more structure’ involving the rankings of preference rankings, which follows Frankfurt (1971). Sen’s meta-ranking entails two sets, X and Y (Sen, 1982, p. 100). As reconstructed in Fig. 2, set X includes bundles Bj , where B1 can be five apples for the self and three apples for the group. Set Y includes elements A through Z where each specifies a ranking of Bj . The A ranking may specify B1 preferable to B2 and the B ranking may stipulate the opposite. The set Y is nothing other than a multitude of preference functions which are exclusive. Each actor (ai = a1 , a2 ,. . . , an ) has a meta-ranking (Yi m ) which expresses ‘a particular system of morality’ (Ibid., p. 101). For an , e.g., the appropriate system is Yn m where, e.g., B > K > . . . > H. Insofar as an agent is able to rank the different rankings means that the multitude of preference functions are consistent and commensurable. So, as Jon Elster (Elster, 1986, pp. 11–13) assesses, meta-ranking is the articulation of one general category capable of ordering more particular sets of preferences. David George (1998) contrasts the idea of metapreference to the multiple-self approach. To one interpretation at least, when the preferences subject to ranking involve varied substantive interests, meta-ranking of preference is merely a more sophisticated formulation of the unitary preference function. In this light, Howard Margolis’s (Margolis, 1982) theory is in the tradition of metapreference: He postulates that behavior expresses a compromise between two functions, ‘self-interest’ and ‘social motivation.’ This entails a meta-function to which the two different selves (selfish and altruist) reconcile. In contrast, Sen uses meta-ranking to advance a multiple-self conception of preference where commitment is not even a preference which can be reconciled with welfare. Nonetheless, Sen’s employment of the meta-ranking lexicon cannot deliver the blow that Sen wants to inflict on conventional theory. Sen somewhat recognizes the shortcoming when he calls
382
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Fig. 3. Reconstruction of Sen’s multiple-self view.
meta-ranking a ‘general technique’ (Sen, 1982, p. 101). That is, meta-ranking is not a specific technique to model commitment. To wit, an can apply any particular moral system to rank the ranking elements (Sen, 1982, p. 100–101). 5 To take a step further, D includes the commitment to some particular group, say to the neighborhood or to the social class to which he belongs. To Sen, the ‘most moral’ ranking M, where he uses ‘M’ in the sense of a particular meta-ranking (Yn m ), can be any of the ranking of A, B, D, and so one. That is, a1 (whose M1 = Y1 m ) may rank B (isolated interests) preferable to D (commitment to group), while agent a2 (whose M1 = Y2 m ) ranks them in reverse. There is nothing in the idea of meta-ranking which entails anything about a conflict between commitment and interest (or ‘preference’ in Sen’s lexicon). The agent can simply change his meta-ranking as he pleases. Thus, the idea of meta-ranking fits like hand in glove with the traditional notion of commensurable preference functions. Sen extricates himself from this traditional implication to assert the thesis that commitment is counter-preferential by making a subtle confusion. He suddenly states, following his definition of M as a meta-ranking (Yi m ), that M can also be defined as an element of the set Y as if the two uses are logically equivalent (Sen, 1982, p. 101). Obviously, the class cannot be a member of itself: M cannot be Yi m and simultaneously an element in the Y set. Yi m , or what Sen calls the ‘most moral’ ranking M,‘ is how to arrange the ranking elements A through Z and, hence, Yi m cannot logically be ‘some other ranking quite distinct’ from the others. Once one considers Yi m (the ‘most moral’) an element in the set Y, the set Y stops being about meta-ranking preference function. It becomes a single-ranking preference function where Bj are arranged according to one criterion: from the less moral B and A elements to the most moral element M. Sen justifies what amounts to the abandonment of meta-ranking and adoption of single-ranking by asserting that ‘moral system requires sacrifice of some self-interest.’ This amounts to admitting that the meta-ranking lexicon cannot capture Sen’s thesis, viz., the moral system requires sacrifice (Sen, 1982, p. 101). In this single-ranking, Sen views choice as a compromise between less desirable preference functions and the ideal ‘most moral’ ranking (Sen, 1982, p. 100).This entails a multiple-self view, as Fig. 3 shows. Sen’s agenda is not about the meta-ranking of preference functions which would approximate the unitary-self approach. It is rather about the inherent opposition between the ‘most moral’ commitments and substantive interests (welfare). So, Sen’s view does not depart from Harsanyi’s ‘dual structure.’ Sen merely adds intermediate selves, ones infused with sympathy (A) and actual choices (C), between the two opposite poles. Thus, Sen’s multiple-self approach is ultimately based on the opposition between interest and moral commitments. Such an opposition stems from the fact that Sen (1987, p. 55n) 5 It is inconsistent to regard C as an element of the set Y (i.e. a separate preference function) as well as a general category representing any other element which is actually acted upon.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
383
ultimately aims at providing the ‘foundation’ of the relevance of basic needs fulfillment. As John Rawls (1988) argues, this amounts to providing a ‘comprehensive doctrine’ or some unique view of the good which is contrary to Rawls’s view of the ‘political conception’ of justice. Notwithstanding Sen’s (Sen, 1990) defense, Sen’s comprehensive doctrine stems from his wanting to ground non-binding commitment, upon which interpersonal preference is conducted, on the solid foundation of binding commitment, i.e., justice. (Sen, 1990, p. 112). The search for such a solid foundation entails the conception of moral canons as stemming from a source separate from substantive preference (Sen, 1980). Such a separation may explain why the multiple-self view cannot ultimately delineate sentimental fools from rational sentimentalists.
7. Origin of sentimental fools Given his multiple-self view, Sen fails to differentiate sentimental fools from rational sentimentalists. For instance, he misread Bernard Shaw’s The Devil’s Disciple. While Shaw exposes the motive of Richard Dudgeon, the protagonist of the play, as an exemplar of sentimental foolishness which characterizes moral dogmatists, Sen celebrates it: Commitment is, of course, closely connected with one’s morals. But moral this question [sic] is in a very broad sense, covering a variety of influences from religious to political, from the ill-understood to the well argued. When, in Bernard Shaw’s The Devil’s Disciple, Judith Anderson interprets Richard Dudgeon’s willingness to be hanged in place of her husband as arising from sympathy for him or love for her, Richard is adamant in his denial: ‘What I did last night, I did in cold blood, caring not half so much for your husband, or for you as I do for myself. I had no motive and no interest: all I can tell you is that when it came to the point whether I would take my neck out of the noose and put another man’s into it, I could not do it’ [Sen, 1982, pp. 93–94]. To substantiate, the play takes place in New England in 1777 as the British troops try to quell the American rebellion. They went to Judith Anderson’s house to arrest her husband, Reverend Anderson, on charges of treason. They mistook Richard as the Reverend. Richard chose not to reveal his identity not because he cares about the Reverend but rather because of a binding commitment. As confirmed by Shaw (1931, p. xxvi) in a section in the preface notably titled ‘On Diabolinian Ethics,’ Richard tells Judith ‘the obvious truth that he would have done as much for any stranger that the law of his own nature, and no interest nor lust whatsoever, forbad him to cry out that the hangman’s noose should be taken off his neck only to be put on another man’s.’ Richard acquired the ‘devil’s disciple’ title for making mockery of pious, but hypocritical Puritans. Richard revelled the persona of the iconoclast who does not lead a mendacious life. He desired to show that he is more moral than the false Puritans who surround him. 6 This desire blinded him as he came to realize later. To complete the sentence from which Sen takes his quote, Richard did not know why he did not reveal his identity, which would not endanger the Reverend’s life. He confessed: ‘I see myself as a fool for my pains’ (Shaw, 1931, p. 52). Richard viewed his life as a failure: He failed to achieve some kind of moral 6
For a general background on Shaw and psychoanalysis, see Arthur Nethercot (1969) and Sidney Albert (1971).
384
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
purity and saw the chance of dying in place of the Reverend as an opportunity to redeem himself. The opportunity blinded him from realizing that the declared purpose viz., saving the Reverend’s life, can be achieved without endangering his own. Richard recognized his sentimental foolishness later when Reverend Anderson was able to save Richard’s life without endangering his own. At the end, Richard stated that he ‘behaved like a fool’ [Ibid. p. 74]. Judith answered that he acted “like a hero.” Richard replied: “much the same thing, perhaps.” With some bitterness towards himself, however, Richard retracts his statement: “But no: if I had been any good, I should have done for you [Reverend Anderson] what you did for me, instead of making a vain sacrifice.” So, Richard’s sacrifice did not arise from rational sentimentality, where benefit is evaluated in light of its cost. At least for Shaw, Richard is a sentimental fool, a Puritan who treated non-binding commitment (care about the welfare of others) as binding, i.e., as oblivious to cost. It is not surprising that Sen, who failed to distinguish the two commitments in relation to cost, also failed to detect Richard’s sentimental or anxiety-driven foolishness.
8. Conclusions The paper is organized around four interrelated points. First, non-binding differs from binding commitment in many ways, especially how each relates to substantive interest. Second, the lumping of non-binding with binding commitments entails that non-binding commitment, similar to binding one, is contrary to interest. Third, such lumping leads, à la Sen, to a multiple-self view which radically separates commitment per se from interest. Fourth, such a view makes it impossible to distinguish sentimental fools from rational sentimentalists. To distinguish the two, we need to distinguish the two kinds of commitment and how each relates to interest, which need not necessarily invoke the unitary-self view. While the paper employs Sen’s preference/choice divergence thesis, it criticizes Sen’s conception of commitment as radically separate from substantive interest. The paper argues that commitment should be considered as a special kind of preference. While commitment is not reducible to interest à la unitary-self view, commitment is contingent on interest contrary to the multiple-self view. The paper proposes two variants of this contingency. First, commitment is non-binding when the agent pursues forward-looking interest which would test his ability. Second, commitment is binding when the agent fulfills a past promise which does not concern the testing of ability. Thus, non-binding commitment, such as the pursuit of a career, and binding commitment, such as honesty, relate differently to substantive interest. The paper shows how moral dogmatism or Puritanism arises when one treats non-binding principles as binding ones. While observance of the rules of justice is absolute, principles like urging higher self-fulfilling goals, kindness, loyalty, group solidarity, the aversion of divorce and abortion, and altruism are subject to deliberation in light of other competing demands. In his useful argument that law is not simply a system of absolute rules, Ronald Dworkin (1977) maintains that principles are wide-ranging and cannot be followed in an absolute manner. In fact, most principles involve conflicts between two goods or two claims, which demands one to compare and apply the cost-benefit calculations. This is not the case with rights. The payment of what one owes the grocer does not involve deciding among
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
385
countervailing interests. It rather arises from a binding commitment which counters, at least in single-shot games, substantive interest. The paper portrays a picture of the archeology of the self which is less simplistic than the standard one. However, the picture is a coherent one because it represents commitments as preferences and, in turn, relates them to substantive interest. While such a parsimonious representation conserves the aesthetic quality which typifies the unitary-self approach, it avoids a number of anomalies which faces the approach such as the origins of embarrassment and shame, the quest after justice even beyond what is justified by deterrence, the inexplicable decline of supply of internal organs once their market is legalized, and so on. Such anomalies would make standard economic theory vulnerable to attacks from other branches of social science armed with non-parsimonious, multiple-self conceptions.
Acknowledgements The paper was supported by research fellowships from the Alexandre von Humboldt Foundation (Germany) and the Earhart Foundation and benefitted from the readings of Amartya Sen, Howard Margolis, Cass Sunstein, anonymous refrees, and especially Steven Joyce. The technical help of Carole Brown is acknowledged. The usual caveat applies. References Albert, Sidney P., 1971. Reflections on Shaw and psychoanalysis. Modern Drama 14 (2), 169–194. Arrow, Kenneth J., 1997. Invaluable goods. Journal of Economic Literature 35 (2), 757–765. Becker, Gary S., 1996. Accounting for Tastes. Harvard University Press, Cambridge, MA. Davidson, D., 1980. How is Weakness of Will Possible? In: Davidson, D., Essays on Actions and Events, ch. 2. Oxford University Press, Oxford. Dworkin, Ronald M., 1977. Is law a System of Rules? In: Dworkin R.M. (Ed.), The Philosophy of Law. Oxford University Press, Oxford. Elster, Jon, 1983. Sour Grapes: Studies in the Subversion of Rationality. Cambridge University Press, Cambridge. Elster, Jon (Ed.), 1986. The Multiple Self. Cambridge University Press, Cambridge. Etzioni, Amitai, 1986. The case for a multiple-preference conception. Economics and Philosophy 2 (2), 159–183. Frankfurt, Harry M., 1971. Freedom of the will and the concept of a person. Philosophical Review 68 (1), 5–20. George, David, 1993. Does the market create preferred preferences?. Review of Social Economy 51 (3), 323–346. George, David, 1998. Coping rationally with unpreferred preferences. Eastern Economic Journaling 24 (2), 181– 194. Hahn, Frank, 1982. One Some Difficulties of the Utilitarian Economist. In: Sen, Amatya, Williams, Bernard (Eds.), Utilitarianism and Beyond. Cambridge University Press, Cambridge, pp. 187–198. Harsanyi, John C., 1955. Cardinal welfare, individualistic ethics, and international comparisons of preference. Journal of Political Economy 63, 309–321. Harsanyi, John C., 1986. Individual Utilities and Utilitarian Ethics. In: Diekmann, Andreas, Mitter, Peter (Eds.), Paradoxical Effects of Social Behavior: Essays in Honor of Anatol Rapoport. Physica-Verlag, Heidelberg, pp. 1–12. Harsanyi, John C., 1986. Rational Behavior and Bargaining Equilibrium in Games and Social Situations. Cambridge University Press, Cambridge. Hausman, Daniel M., McPherson, Michael S., 1993. Taking ethics seriously: Economics and contemporary moral philosophy. Journal of Economic Literature 31 (2), 671–731. Jeffrey, R.C., 1974. Preferences Among Preferences, The Journal of Philosophy 377–391.
386
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Kavka, Gregory S., 1991. Is individual choice less problematic than collective choice?. Economics and Philosophy 7 (2), 143–165. Khalil, Elias L., 1995. On the scope of economics: What is the question?. Finnish Economic Papers 8 (1), 38–53. Khalil, Elias L., 1996. What is economic action? From Marshall and Robbins to Polanyi and Becker. Journal of the History of Economic Thought 18 (1), 13–36. Khalil, Elias L., 1997a. Buridan’s ass, risk, uncertainty, and self-competition: A theory of entrepreneurship. Kyklos 50 (2), 147–163. Khalil, Elias L., 1997b. Etzioni versus Becker: Do moral sentiments differ from ordinary tastes?. De Economist 145 (4), 491–520. Khalil, Elias L., 1999. Symbolic Products: Prestige, Pride and Identity Goods. Theory and Decision, in press. Kuran, Timur, 1995. Private Truths, Public Lies: The Social Consequences of Preference Falsification. Harvard University Press, Cambrdige, MA. Lewis, Helen Block, 1971. Shame and Guilt in Neurosis. International Universities Press, New York. Margolis, Howard, 1982. Selfishness, Altruism and Rationality: A Theory of Social Choice, Cambridge University Press, Cambridge. McPherson, Michael S., 1980. Liberty and the higher pleasures. In defense of mill. Economic Inquiry 8 (2), 314–320. Mueller, D., 1986. Rational egoism vs. adaptive egoism. Public Choice 51, 3–23. Nethercot, Arthur H., 1969. Bernard Shaw and psychoanalysis. Modern Drama 11 (4), 356–375. Posner, Richard A., 1995. Aging and Old Age. University of Chicago Press, Chicago. Radin, Margaret Jane, 1996. Contested Commodities. Harvard University Press, Cambridge, MA. Rawls, John;, 1971. A Theory of Justice. Harvard University Press, Cambridge, MA. Rawls, John, 1988. The Priority of right and ideas of the good. Philosophy and Public Affairs 17 (4), 251–276. Schelling, Thomas C., 1984b. Choice and Consequence. Harvard University Press, Cambridge, MA. Sen, Amartya K., 1977. Rational fools: A critique of the behavioral foundations of economic theory (Reprinted in Sen, 1982, ch. 4). Philosophy & Public Affairs) 6 (4), 317–344. Sen, Amartya K., 1980-1981. Plural Preference. Proceedings of the Aristotelian Society, pp. 193–215. Sen, Amartya K., 1982. Choice, Welfare and Measurement. Basil Blackwell, Oxford. Sen, Amartya K., 1985. Goals, commitment, and identity. Journal of Law, Economics and Organization 1 (2), 341–355. Sen, Amartya K., 1987. On Ethics and Economics. Basil Blackwell, Oxford. Sen, Amartya K., 1990. Justice: Means versus freedoms. Philosophy and Public Affairs 19 (2), 111–121. Sen, Amartya K., 1993. Internal consistency of choice. Econometrica 61 (3), 495–521. Sen, Amartya K., 1995. Is the Idea of Purely Internal Consistency of Choice Bizarre? In: Altham, J.E.J., Harrison, Ross (Eds.), World, Mind and Ethics: Essays on the Ethical Philosophy of Bernard Williams. Cambridge University Press, Cambridge, pp. 19–31. Sen, Amartya K., 1997. Maximization and the act of choice. Econometrica 65 (4), 745–779. Shaw, Bernard,1931. Three Plays for Puritans: The Devil’s Disciple, Caesar and Cleopatra, and Captain Brassbound’s Conversion. Constable and Company, London, 1900. Stewart, Hamish, 1992. Rationality and the market for human blood. Journal of Economic Behavior and Organization 19 (2), 125–143. Sunstein, Cass R., 1997. Free Markets and Social Justice. Oxford University Press, Oxford. Taylor, Charles, 1995. A most peculiar institution. In: Altham, J.E.J., Harrison, Ross (Eds.), World, Mind and Ethics: Essays on the Ethical Philosophy of Bernard Williams. Cambridge University Press, Cambridge, pp. 132–155.
Sentimental fools: a critique of Amartya Sen’s notion of commitment Elias L. Khalil ∗ 1338 F. Hyde Park Blvd., #3, Chicago, IL 60615, USA Received 25 October 1997; received in revised form 8 February 1999; accepted 11 March 1999
Abstract Commitment is problematic because one sometimes pursues it against one’s interest. To solve it, the paper proposes a distinction between ‘non-binding’ and ‘binding’ commitments. Non-binding commitment is about ambition, such as becoming a great chef, which bolsters welfare in the pecuniary sense as well as self-respect. In contrast, ‘binding commitment’ is about honesty. While it diminishes welfare, it augments self-integrity. The neoclassical view reduces both commitments to interest, while the multiple-self approach separates both commitments from interest. The separation permits the confusion of sentimental fools, who enter commitments without regard to interest, with rational sentimentalists, who take interest into consideration. ©1999 Elsevier Science B.V. All rights reserved. JEL classification: D0 Keywords: Substantive interest; Non-binding commitment; Binding commitment; Sentimental fools; Rational fools; Rational sentimentalists; Guilt; Disgrace (shame); Embarrassment; Substantive preference; Symbolic preference (self-respect and self-integrity); Lethargic will; Crooked will
1. Introduction Given that commitment appears to negate interest, some theorists, such as Amartya Sen, propose that agents have conflicting selves. Defenders of the standard approach, on the other hand, deny the existence of the problem; they simply regard commitment as an ordinary preference stemming from a unitary self. While this paper recognizes the phenomenon of commitment as a problem, it tries to solve it without appealing to the multiple-self view. ∗ Tel.: +1-773-548-7638; fax: +1-773-702-8490 E-mail address: [email protected] (E.L. Khalil)
0167-2681/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 2 6 8 1 ( 9 9 ) 0 0 0 6 0 - 8
374
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Fig. 1. The three preferences.
It does so by distinguishing two kinds of commitment and how each relates differently to interest–a necessary distinction to discern admirable acts from sentimental foolishness. Let us start with three kinds of preferences, which may overlap, facing the self (S): Situation 1: S decides to enter a training program in the culinary art of French cuisine because the going wage rate for chefs, given the cost of training and the responsibility of the job, is more attractive than other options. Situation 2: In addition to Situation 1, S has been impressed by the creative art involved in French cooking. He takes on a commitment to try to become the best chef in the city. Situation 3: In addition to Situation 2, S takes on the commitment to be truthful. In specific, he prepares, in light of his advertisement, dishes made only from fresh ingredients, even when the pecuniary benefit from a cheating strategy is unambiguously greater than expected pecuniary cost resulting possibly from a tarnished market reputation. Situation 1 involves what I call ‘substantive interest’, pecuniary calculation. Situation 2 consists of ‘non-binding commitment’ because the promise cannot be obligatory given that the agent is uncertain whether he can fulfill it. Situation 3 entails ‘binding commitment’ because the agent is certain that he can meet his obligation. The agent fulfills his binding commitment to enjoy self-integrity. If he fulfills it exclusively out of substantive calculation concerning penalty, it would fail to generate the sense of self-integrity. While Situation 2 advances substantive preference, Situation 3 by definition reduces it. Nonetheless, both kinds of commitment satisfy a sense of selfhood, dubbed here ‘symbolic preference’ as opposed to ‘substantive preference’. As summed up in Fig. 1, while the negation of one’s substantive interest would be irrational, the negation of either kind of commitment is usually described as the result of weak will. There are two kinds of weak will (‘lethargic’ and ‘crooked’), and two kinds of symbolic preference (‘self-respect’ and ‘self-integrity’), following the two kinds of commitment. The paper commences with some motivation (Section 1). It distinguishes binding from non-binding commitments (Section 2) and preference from choice (Section 3). Many authors confuse non-binding with binding commitment, which lead them to separate commitment per se from substantive interest (Section 4). Such a separation entails that the agent is constituted of multiple selves (Section 5). Such a view, in turn, makes it impossible to distinguish sentimental fools from rational sentimentalists (Section 6).
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
375
2. Motivation While the three situations are about self-concern, one can equally illustrate them by other-concerns such as altruism, public spiritedness, or patriotism. The conjectured distinction among the three preferences has two payoffs, practical and theoretical. Regarding the practical payoff, the debate about income redistribution can be clarified on the ground of whether it is prompted by altruism, which pertains to substantive interest, or is instigated by fairness, which involves binding commitment. If a redistribution policy is prompted by altruism, one has to weigh the efficiency advantage versus side-effect disadvantages such as the weakening of incentives. But if the policy is prompted by fairness, on the basis of some social contract entered under the veil of ignorance (Rawls, 1971), one would conceive differently the so-called weakening of incentives: It would be in one’s right, as specified by the social contract, not to work below a guaranteed social wage. If there is no difference between commitment and interest, the judicial system’s main goal would be the deterrence of crime rather than the vindication of justice. Therefore, it would be cheaper to punish rich criminals with high fines rather than with jail sentences–as long as both methods are equivalent in deterring crime. However, people usually abhor the idea of fining rich criminals rather than putting them in jail. The abhorrence is inexplicable if one views commitment as a substantive preference. Moreover, the distinction between the two kinds of commitment will make it clear, as shown below, why the negation of non-binding commitment is only a cause for embarrassment, while the negation of binding commitment is usually seen as shameful. In fact, the distinction would provide a ground for why laws commanding charity, as distinct from upholding justice, can be counterproductive. In addition, if one views internal organs, such as kidneys, to be ‘gifts of life’, one would regard the legalization of their trade as repugnant, no different from prostitution (Radin, 1996; cf. Arrow, 1997). One would see the legalization of the market for organs as a violation of binding commitments which the agent, although implicitly, makes with one’s sense of identity. Some experts argue that legalization of the organs market would actually decrease the total organs available similarly to how the legalization of blood donation in the US has hindered the supply of blood (Stewart, 1992). While substantive preference rises as a result of the legalization, it may not rise enough to offset the weakening of symbolic preference which acts as a motive in institutions where the marketing of organs is prohibited. Regarding the theoretical payoff, the focus here, the proposed three-way distinction promises to supersede the shortcomings of both the unitary-self and multiple-self approaches. 1 The standard, unitary-self view does not recognize the interest/commitment distinction. When commitments are not the outcome of strategic repetitive games, the standard approach treats them as preferences reducible to substantive preference function which 1 I use the term ‘multiple-self’ to express inconsistent preferences, where the agent is assumed to contain a ‘moral self’ as opposed to ‘material self.’ In this sense, my usage differs from Richard Posner’s (Posner, 1995, ch. 4). Posner takes the term to denote conflicting needs, where agents are seen as trying to accommodate pressing needs arising from work, family, church, and so on (also see Kuran, 1995, pp. 43–44]. In the sense of inconsistent preferences, the multiple-self view is advanced by Amartya Sen (1982), John Harsanyi (1955), Thomas Schelling (1984), Amitai Etzioni (1986), Gregory Kavka (1991), and Jon Elster (Elster, 1986; cf. Hausman and McPherson, 1993). It is beyond the scope here to contrast fully the multiple- and unitary-self approaches (Khalil, 1995, 1996, 1997b, 1998).
376
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
is the object of optimization. On the other hand, the multiple-self approach recognizes the interest/commitment difference. But it treats non-binding commitment as not different from binding commitment. The treatment entails that commitment stems from a self separate from the one which expresses substantive interest. Such a view, as detailed below with respect to Sen’s, makes it impossible to distinguish sentimental fools from rational sentimentalists. The paper evaluates the advantages of the proposed three-way distinction vis-à-vis Sen’s work. In his 1977 celebrated paper, Sen [(Sen, 1977) (reprinted 1982, ch. 4] argues that the reduction of commitment to interest implies that agents are ‘rational fools’ because they are ready, in response to change of relative price or income, to cheat without the feeling of shame (disgrace) and to quit projects at the first appearance of difficulty without the feeling of embarrassment. It is true that casual observations of normal agents who occasionally cheat and, as soon as the project becomes harder, quit indicate that they experience shame and embarrassment--an experience alien to the rational fools inhabiting the unitary-self world. However, Sen’s separation of commitment from interest invites from the rear window another kind of fools, what is coined here ‘sentimental fools’. Sentimental fools are ready to incur costs unjustified by expected benefits, such as jumping off a ship in the middle of the ocean to retrieve a toy for a crying child. Sen is correct that the collapsing of commitment with interest engenders a world of rational fools, but his radical separation of interest and commitment engenders a world of sentimental fools. To avoid rational fools à la unitary-self view and sentimental fools à la multiple-self view, we should neither collapse interest and commitment nor separate them radically. It is possible to achieve this task with the help of the proposed distinction between non-binding and binding commitments.
3. The commitment dichotomy The difference between the two kinds of commitment is not that binding commitment is coerced, while the other is optional. Both are voluntary. No one coerces S in Situation 3 to advertize his fresh ingredients. Moreover, the difference between the two is not that binding commitment is a promise to a second party, while the other is a personal promise. Binding commitment can be made with the self, as when one refuses to accept payment for blood donation or sexual pleasure. Also, the issue is not that binding commitment is a non-binding one made in the past. All non-binding commitments move into the past once they are made. Rather, the main difference lies with regard to ability. In Situation 1, S is maximizing substantive preference which is assumed to be within one’s ability. In Situation 2, S is also pursuing substantive preference, but not in the strict optimization sense. The goal is rather expressive of S’s self-image, an image to which S makes a non-binding commitment. The commitment is non-binding because S is uncertain whether S has the ability to sustain the necessary effort to carry it through. S may eventually have to give up the commitment if it impairs sufficiently his substantive interest. However, insofar as S acts to attain such a goal, it affords S symbolic preference which goes beyond substantive preference. In contrast, in Situation 3, S is obliged by a binding commitment–pursued even in single-shot games–not to cheat or disown one’s identity, which can be extended to be loyal to one’s family, country,
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
377
tribe, and religion. The commitment is binding because S is certain that S has the ability to sustain the necessary effort to carry it through. In other words, while the agent’s ability is not in question in binding commitment, it is so in non-binding commitment (Khalil, 1997a). In binding commitment, the agent promises to be honest, not to sell or buy sexual pleasure (prostitution), not to accept money for donated blood, not to accept bribes, or to sell his country’s secrets to the enemy when it is clear that such a promise is within his ability. Given that ability is not in doubt, the reason behind the commitment, therefore, is to counter the opportunistic gain which would arise from cheating. In contrast, in non-binding commitment, the agent promises to pursue a career or advance a public cause despite anticipated arduous hurdles. The promise is needed in order to sustain one’s effort given that the ability of the actor with regard to the challenge is uncertain. This situation is similar to the commitment to climb a mountain, whose height and terrain are quite known to the actor, but would be an uncertain challenge for him because he has not tried it before. If the actor quits after trying hard enough, the actor would not be labeled (by himself or others) a ‘quitter’, i.e., someone who lacks fortitude. In contrast, the actor would be labeled ‘corrupt’ if he, as long as he is not in a state of destitute, accepts a bribe irrespective of its size. While the commitment to climb the mountain is contingent on an uncertain ability and hence is non-binding, the commitment to rebuff bribes is not contingent on ability and hence is binding. Further, each kind of commitment relates differently to substantive interest. While Situation 1 and Situation 2 augment substantive preference, Situation 2 additionally attains symbolic preference, in specific ‘self-respect’, for exhibiting tenacity. In contrast, Situation 3 originates symbolic preference, in specific ‘self-integrity’, for acting honestly, but only at the expense of lowering substantive preference as clearly in the case of single-shot games. 2 One may appeal to unreflected habit (Mueller, 1986) in explaining why one would act honestly in single-shot games. However, such an appeal is an option of last resort because it invokes ad hoc reasoning. A better explanation is that the agent in Situation 3 is optimizing a function which includes, besides the familiar substantive preference, self-integrity. While self-integrity arises from honesty which may lower substantive interest, self-respect stems from pursuing an ambition which enhances substantive interest. Another way to distinguish between the two commitments is to discern two kinds of weak will or temptations. When one gives up the pursuit of a non-binding commitment at the first sign of difficulty, as in Situation 2, his will can be characterized as ‘lethargic’. This differs from violating binding commitment, as cheating customers in Situation 3, where the will can be depicted as ‘crooked’. While lethargic will engenders the feeling of ‘embarrassment’, crooked will gives rise to the feeling of ‘shame’ for the simple reason that lethargic will invites complacency while crooked will is about duplicity. One can, with respect to binding commitment, be characterized by a crooked will while, 2 The proposed term ‘symbolic preference’ resembles Timur Kuran’s (Kuran, 1995) ‘expressive utility’ (see Khalil, 1999). There is one difference, however. The term ‘symbolic’ is merely an umbrella category, while Kuran’s ‘expressive’ is a concept which supposedly denotes a single preference function. To wit, Kuran, similar to many other writers on the subject, does not distinguish between the two radically different kinds of commitment and the consequently different kinds of symbolic preference. Aside from lexicon, symbolic preference is a category rather than a concept which denotes a phenomenon.
378
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
with regard to non-binding commitment, be characterized by a non-lethargic, tenacious will. 3 Anxiety-driven actions which may indicate neurosis (see Lewis, 1971) are motivated by guilt. Such actions may fail to highlight the proposed difference between the two kinds of weak will. The term ‘guilt’, as used below, exposes situations where action is motivated by anxiety rather than by preference. The term denotes suboptimal preference which one may choose out of lack of self-assertion, boldness, or self-confidence. As Charles Taylor (1995, pp. 142) argues, turning away from the crushing burden of medieval Catholic notion of ‘sin’ or, in contemporary vernacular, ‘guilt’, amounts to liberation: “To break with a good to which one cannot really subscribe is a liberation in anyone’s language.” There are other ways to establish the commitment dichotomy. At least concerning the three ways mentioned so far, they do not provide a clear-cut blueprint on how to classify unambiguously actual preferences. Many acts, such as volunteering resources to a homeless shelter, can be composite ones, blending the motives of altruism (i.e. substantive interest) with ambition (i.e. non-binding commitment to take pride in one’s neighborhood) and justice (i.e. binding commitment to the unlucky ones in society). At least theoretically, however, it is possible to discern the three preferences and two kinds of commitment they entail.
4. The divergence of preference and choice Among others, Becker (1996), Kuran (1995), and Sunstein (1997) propose other reasons, beside guilt-driven anxiety, behind the preference/choice divergence. To note, the set of concepts offered by each author is geared to solve a particular problem, rather than to solve the preference/choice divergence per se. For example, Becker, with other co-authors, tries to account for addiction, where agents choose bundles which are contrary to their preferences. He reasons that the preference/choice divergence is only apparent: Addiction, similar to habits, is the outcome of rational choice when consumption is intertemporally nonseparable. His account implies that observed preference/choice divergence is really the divergence of choice from, what I dub here, ‘fanciful preference’. In the case of addiction, the addict may wish to quit, but his preference might be fanciful when, given the hardship of quitting, it is beyond his budget constraint. The preference for a drug-free life style might be as fanciful as the dream of a school teacher with limited income to own a mansion in the French Riviera. Becker’s theory should not be misunderstood as stating that it is impossible to quit addiction or to live in a mansion with servants. Rather, to attain such objectives, the addict or the school teacher must expend enormous cost to cultivate appropriate new habits, such as socializing with non-addicts or augmenting additional human capital. Otherwise, to close the gap between choice and fanciful preference, the agent must adjust the fanciful preference as he learns from information. Further, Becker’s theory should not be misunderstood as failing to recognize the possibility of pathology, the persistence of the gap between choice and fanciful preference. But such persistence is the result of the misfunctioning of information processing. So, the 3
The analyses of Davidson (1980) and Jon Elster (1983) mostly focus on crooked will.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
379
persistence of the gap is reasoned to be the outcome of special characteristic of information, not expressive of a gap between choice and feasible preference. Becker’s reasoning shows that if one wants to assert the preference/choice divergence thesis, one must first treat preference as feasible. This seems to be the case with Frankfurt’s (Frankfurt, 1971) notion of ‘free will’ with regard to ‘second order preference’ or what came to be called ‘metapreference’ (also see Jeffrey, 1974; McPherson, 1980; Hahn, 1982). One may prefer choosing drugs over education, while one simultaneously prefers the preference for education over drugs at the meta level. For such a metapreference to be non-trivial, George (1993) explicitly distinguishes metapreference from ‘regret’ which is associated with particular choices. In such choices, regret express situations where alternative options are fanciful, i.e., beyond one’s budget constraint. On the other hand, Kuran studies the preference/choice divergence with a focus on the dynamics between private preferences and hostile public opinion. He argues that agents may not act on their private preferences if they do not jibe with public preferences. Kuran’s main objective is not to solve the preference/choice divergence in general. Rather, he tries to account for abrupt changes in public preferences. To Kuran, agents may ‘live a lie’ in public, falsify their private preferences to secure the approval of others. Such ‘preference falsification’ is justified i.e., not judged as indicative of insincerity or dishonesty only when the disapproval of the public involves severe persecution. For Kuran’s purpose, the preference/choice divergence is relevant only insofar as it sheds light on the divergence of private preference from public choice. Thus, this particular divergence stems from historical accidents of coercive social-political environments. In the absence of such an environment, and especially when public opinion values maverick and eccentric private preferences, there is no need for public choice to diverge from private preference. Nonetheless, Kuran’s thesis is useful here. It clarifies one issue. If one wants to assert the preference/choice divergence thesis, one must be clear that the issue is about truthful choice where the approval of others is irrelevant. Further, Sunstein investigates the preference/choice divergence with a focus on how societal corrupting norms may push choice away from true preference. He maintains that preferences are fuzzy and, hence, susceptible to the manipulation of unenlightened societal norms such as the pursuit of vanity. This entails that there is a need for state intervention in order to align choices with true preferences. Sunstein’s thesis about the deviation of choice from true preference does not fully coincide with the preference/choice deviation discussed here. For one thing, there would be no divergence if the public is fully enlightened even to the satisfaction of the benevolent state. Nonetheless, Sunstein’s discussion clarifies one issue. If one wants to assert the preference/choice divergence thesis, one must be clear that the issue is about the divergence of choice from enlightened preference.
5. Lumping non-binding with binding commitments Sen (1982, p. 1; 1985, 1993, 1995) proposes the preference/choice divergence from a different angle. He argues, contrary to the axioms of the theory of revealed preference, one should not assume a priori that behavior is internally consistent. Empirical study should rather determine whether, in a particular case, choice matches preference. While the
380
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
paper does not quibble with this thesis, Sen extends the thesis to imply that the agent has multiple selves. This implication is unwarranted. To show this, we need to unpack Sen’s confusing categories. The confusion arises from a terminological twist which equates the preference/choice divergence with the interest/commitment distinction. He reaches such an identification by equating the terms ‘preference’ with ‘interest’, on one hand, and ‘choice’ with ‘commitment’, on the other. While the issue is not about lexicon, this paper employs ‘preference’ to denote both interest and commitment, and utilizes ‘choice’ to indicate action which may deviate from preference. When Sen equates preference with interest and choice with commitment, he can claim with such a terminological twist that the interest/commitment distinction is necessarily implied by the accepted preference/choice divergence. With this lexicon, Sen overlooks the commitment dichotomy when he lumps both kinds of commitment in one concept and, in turn, opposes it to interest as if each kind of commitment relates in the same way to interest. For instance, he explicitly opposes ‘commitment’ and ‘sympathy’. He illustrates it by distinguishing two separate motives behind the opposition to torture: one is out of principle and the other is out of sympathy in the sense of caring about the substantive interest of the other (Sen, 1982, pp. 91–92). He further illustrates the difference: The contrast between sympathy and commitment may be illustrated with the story of two boys who find two apples, one large, one small. Boy A tells boy B, ‘You choose’. B immediately picks the larger apple. A is upset and permits himself the remark that this was grossly unfair. ‘Why?’ asks B. ‘Which one would you have chosen, if you were to choose rather than me?’ ‘The smaller one, of course’, A replies. B is now triumphant: ‘Then what are you complaining about? That’s the one you’ve got!’ B certainly wins this round of argument, but in fact A would have lost nothing from B’s choice had his own hypothetical choice of the smaller apple been based on sympathy as opposed to commitment. A’s anger indicates that this was probably not the case (Sen, 1982, p. 93) also see (Sen, 1993, p. 501). 4 It is clear that Sen places ‘sympathy’ in the substantive interest (welfare) function, while he treats ‘commitment’ as a category which reduces that function. Sen’s interest/commitment opposition makes sense if the word ‘commitment’ strictly denotes binding commitment. However, Sen uses the term to also denote non-binding commitment such as the commitment ‘to exert’ oneself, which rather augments rather than reduces substantive interest (Sen, 1982, pp. 97–98). Sen uses the same term to discuss the binding rule of justice to abstain from the use of torture and to express striving towards challenging goals, as working harder for a better future. Sen must have treated non-binding commitment as binding when he argues in many places that commitment per se always lowers substantive interest (welfare) (e.g. Sen, 1982, p. 93). For Sen, ‘commitment does involve, in a very real sense, counterpreferential choice’ [Ibid.]. Such an interest/commitment opposition overlooks that non-binding commitment 4 Sen (1997, p. 747) repeats the same observation with regard to the tendency of an agent not to grab immediately his favorite chair in a party or snatch his favorite fruit from the host’s tray. Sen dismisses the view that such etiquettes are ways which allow an actor to sympathize with the intensity of preferences of others [Ibid., p. 747n]. He maintains that it arises either from the concern over reputation, social commitment to a moral imperative, or conventional rule following.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
381
Fig. 2. Sen’s meta-ranking.
buttresses interest. As discussed next, the opposition leads to a particular view of ranking of preferences, the multiple-self view. As shown later, such a view cannot afford the conceptual tools needed to distinguish sentimental fools from rational sentimentalists.
6. Meta-ranking of preferences and the multiple-self view Sen welcomes John Harsanyi’s (Harsanyi, 1955, 1986, 1988) distinction between ‘ethical’ and ‘subjective’ preferences. While the former expresses what the person prefers ‘on the basis of impersonal social considerations alone,’ the latter denotes what the person actually prefers ‘on the basis of his personal interests or on any other basis’ [Ibid., p. 315]. This sociological approach generally defines the ‘good’ from the social rather than the individual standpoint. For Sen (1982, p. 100), however, Harsanyi’s ‘dual structure is deficient.’ First, ‘actual’ behavior may express ethical consideration and, second, ‘a preference ordering can be more ethical than another but less so than a third.’ Sen proposes ‘more structure’ involving the rankings of preference rankings, which follows Frankfurt (1971). Sen’s meta-ranking entails two sets, X and Y (Sen, 1982, p. 100). As reconstructed in Fig. 2, set X includes bundles Bj , where B1 can be five apples for the self and three apples for the group. Set Y includes elements A through Z where each specifies a ranking of Bj . The A ranking may specify B1 preferable to B2 and the B ranking may stipulate the opposite. The set Y is nothing other than a multitude of preference functions which are exclusive. Each actor (ai = a1 , a2 ,. . . , an ) has a meta-ranking (Yi m ) which expresses ‘a particular system of morality’ (Ibid., p. 101). For an , e.g., the appropriate system is Yn m where, e.g., B > K > . . . > H. Insofar as an agent is able to rank the different rankings means that the multitude of preference functions are consistent and commensurable. So, as Jon Elster (Elster, 1986, pp. 11–13) assesses, meta-ranking is the articulation of one general category capable of ordering more particular sets of preferences. David George (1998) contrasts the idea of metapreference to the multiple-self approach. To one interpretation at least, when the preferences subject to ranking involve varied substantive interests, meta-ranking of preference is merely a more sophisticated formulation of the unitary preference function. In this light, Howard Margolis’s (Margolis, 1982) theory is in the tradition of metapreference: He postulates that behavior expresses a compromise between two functions, ‘self-interest’ and ‘social motivation.’ This entails a meta-function to which the two different selves (selfish and altruist) reconcile. In contrast, Sen uses meta-ranking to advance a multiple-self conception of preference where commitment is not even a preference which can be reconciled with welfare. Nonetheless, Sen’s employment of the meta-ranking lexicon cannot deliver the blow that Sen wants to inflict on conventional theory. Sen somewhat recognizes the shortcoming when he calls
382
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Fig. 3. Reconstruction of Sen’s multiple-self view.
meta-ranking a ‘general technique’ (Sen, 1982, p. 101). That is, meta-ranking is not a specific technique to model commitment. To wit, an can apply any particular moral system to rank the ranking elements (Sen, 1982, p. 100–101). 5 To take a step further, D includes the commitment to some particular group, say to the neighborhood or to the social class to which he belongs. To Sen, the ‘most moral’ ranking M, where he uses ‘M’ in the sense of a particular meta-ranking (Yn m ), can be any of the ranking of A, B, D, and so one. That is, a1 (whose M1 = Y1 m ) may rank B (isolated interests) preferable to D (commitment to group), while agent a2 (whose M1 = Y2 m ) ranks them in reverse. There is nothing in the idea of meta-ranking which entails anything about a conflict between commitment and interest (or ‘preference’ in Sen’s lexicon). The agent can simply change his meta-ranking as he pleases. Thus, the idea of meta-ranking fits like hand in glove with the traditional notion of commensurable preference functions. Sen extricates himself from this traditional implication to assert the thesis that commitment is counter-preferential by making a subtle confusion. He suddenly states, following his definition of M as a meta-ranking (Yi m ), that M can also be defined as an element of the set Y as if the two uses are logically equivalent (Sen, 1982, p. 101). Obviously, the class cannot be a member of itself: M cannot be Yi m and simultaneously an element in the Y set. Yi m , or what Sen calls the ‘most moral’ ranking M,‘ is how to arrange the ranking elements A through Z and, hence, Yi m cannot logically be ‘some other ranking quite distinct’ from the others. Once one considers Yi m (the ‘most moral’) an element in the set Y, the set Y stops being about meta-ranking preference function. It becomes a single-ranking preference function where Bj are arranged according to one criterion: from the less moral B and A elements to the most moral element M. Sen justifies what amounts to the abandonment of meta-ranking and adoption of single-ranking by asserting that ‘moral system requires sacrifice of some self-interest.’ This amounts to admitting that the meta-ranking lexicon cannot capture Sen’s thesis, viz., the moral system requires sacrifice (Sen, 1982, p. 101). In this single-ranking, Sen views choice as a compromise between less desirable preference functions and the ideal ‘most moral’ ranking (Sen, 1982, p. 100).This entails a multiple-self view, as Fig. 3 shows. Sen’s agenda is not about the meta-ranking of preference functions which would approximate the unitary-self approach. It is rather about the inherent opposition between the ‘most moral’ commitments and substantive interests (welfare). So, Sen’s view does not depart from Harsanyi’s ‘dual structure.’ Sen merely adds intermediate selves, ones infused with sympathy (A) and actual choices (C), between the two opposite poles. Thus, Sen’s multiple-self approach is ultimately based on the opposition between interest and moral commitments. Such an opposition stems from the fact that Sen (1987, p. 55n) 5 It is inconsistent to regard C as an element of the set Y (i.e. a separate preference function) as well as a general category representing any other element which is actually acted upon.
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
383
ultimately aims at providing the ‘foundation’ of the relevance of basic needs fulfillment. As John Rawls (1988) argues, this amounts to providing a ‘comprehensive doctrine’ or some unique view of the good which is contrary to Rawls’s view of the ‘political conception’ of justice. Notwithstanding Sen’s (Sen, 1990) defense, Sen’s comprehensive doctrine stems from his wanting to ground non-binding commitment, upon which interpersonal preference is conducted, on the solid foundation of binding commitment, i.e., justice. (Sen, 1990, p. 112). The search for such a solid foundation entails the conception of moral canons as stemming from a source separate from substantive preference (Sen, 1980). Such a separation may explain why the multiple-self view cannot ultimately delineate sentimental fools from rational sentimentalists.
7. Origin of sentimental fools Given his multiple-self view, Sen fails to differentiate sentimental fools from rational sentimentalists. For instance, he misread Bernard Shaw’s The Devil’s Disciple. While Shaw exposes the motive of Richard Dudgeon, the protagonist of the play, as an exemplar of sentimental foolishness which characterizes moral dogmatists, Sen celebrates it: Commitment is, of course, closely connected with one’s morals. But moral this question [sic] is in a very broad sense, covering a variety of influences from religious to political, from the ill-understood to the well argued. When, in Bernard Shaw’s The Devil’s Disciple, Judith Anderson interprets Richard Dudgeon’s willingness to be hanged in place of her husband as arising from sympathy for him or love for her, Richard is adamant in his denial: ‘What I did last night, I did in cold blood, caring not half so much for your husband, or for you as I do for myself. I had no motive and no interest: all I can tell you is that when it came to the point whether I would take my neck out of the noose and put another man’s into it, I could not do it’ [Sen, 1982, pp. 93–94]. To substantiate, the play takes place in New England in 1777 as the British troops try to quell the American rebellion. They went to Judith Anderson’s house to arrest her husband, Reverend Anderson, on charges of treason. They mistook Richard as the Reverend. Richard chose not to reveal his identity not because he cares about the Reverend but rather because of a binding commitment. As confirmed by Shaw (1931, p. xxvi) in a section in the preface notably titled ‘On Diabolinian Ethics,’ Richard tells Judith ‘the obvious truth that he would have done as much for any stranger that the law of his own nature, and no interest nor lust whatsoever, forbad him to cry out that the hangman’s noose should be taken off his neck only to be put on another man’s.’ Richard acquired the ‘devil’s disciple’ title for making mockery of pious, but hypocritical Puritans. Richard revelled the persona of the iconoclast who does not lead a mendacious life. He desired to show that he is more moral than the false Puritans who surround him. 6 This desire blinded him as he came to realize later. To complete the sentence from which Sen takes his quote, Richard did not know why he did not reveal his identity, which would not endanger the Reverend’s life. He confessed: ‘I see myself as a fool for my pains’ (Shaw, 1931, p. 52). Richard viewed his life as a failure: He failed to achieve some kind of moral 6
For a general background on Shaw and psychoanalysis, see Arthur Nethercot (1969) and Sidney Albert (1971).
384
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
purity and saw the chance of dying in place of the Reverend as an opportunity to redeem himself. The opportunity blinded him from realizing that the declared purpose viz., saving the Reverend’s life, can be achieved without endangering his own. Richard recognized his sentimental foolishness later when Reverend Anderson was able to save Richard’s life without endangering his own. At the end, Richard stated that he ‘behaved like a fool’ [Ibid. p. 74]. Judith answered that he acted “like a hero.” Richard replied: “much the same thing, perhaps.” With some bitterness towards himself, however, Richard retracts his statement: “But no: if I had been any good, I should have done for you [Reverend Anderson] what you did for me, instead of making a vain sacrifice.” So, Richard’s sacrifice did not arise from rational sentimentality, where benefit is evaluated in light of its cost. At least for Shaw, Richard is a sentimental fool, a Puritan who treated non-binding commitment (care about the welfare of others) as binding, i.e., as oblivious to cost. It is not surprising that Sen, who failed to distinguish the two commitments in relation to cost, also failed to detect Richard’s sentimental or anxiety-driven foolishness.
8. Conclusions The paper is organized around four interrelated points. First, non-binding differs from binding commitment in many ways, especially how each relates to substantive interest. Second, the lumping of non-binding with binding commitments entails that non-binding commitment, similar to binding one, is contrary to interest. Third, such lumping leads, à la Sen, to a multiple-self view which radically separates commitment per se from interest. Fourth, such a view makes it impossible to distinguish sentimental fools from rational sentimentalists. To distinguish the two, we need to distinguish the two kinds of commitment and how each relates to interest, which need not necessarily invoke the unitary-self view. While the paper employs Sen’s preference/choice divergence thesis, it criticizes Sen’s conception of commitment as radically separate from substantive interest. The paper argues that commitment should be considered as a special kind of preference. While commitment is not reducible to interest à la unitary-self view, commitment is contingent on interest contrary to the multiple-self view. The paper proposes two variants of this contingency. First, commitment is non-binding when the agent pursues forward-looking interest which would test his ability. Second, commitment is binding when the agent fulfills a past promise which does not concern the testing of ability. Thus, non-binding commitment, such as the pursuit of a career, and binding commitment, such as honesty, relate differently to substantive interest. The paper shows how moral dogmatism or Puritanism arises when one treats non-binding principles as binding ones. While observance of the rules of justice is absolute, principles like urging higher self-fulfilling goals, kindness, loyalty, group solidarity, the aversion of divorce and abortion, and altruism are subject to deliberation in light of other competing demands. In his useful argument that law is not simply a system of absolute rules, Ronald Dworkin (1977) maintains that principles are wide-ranging and cannot be followed in an absolute manner. In fact, most principles involve conflicts between two goods or two claims, which demands one to compare and apply the cost-benefit calculations. This is not the case with rights. The payment of what one owes the grocer does not involve deciding among
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
385
countervailing interests. It rather arises from a binding commitment which counters, at least in single-shot games, substantive interest. The paper portrays a picture of the archeology of the self which is less simplistic than the standard one. However, the picture is a coherent one because it represents commitments as preferences and, in turn, relates them to substantive interest. While such a parsimonious representation conserves the aesthetic quality which typifies the unitary-self approach, it avoids a number of anomalies which faces the approach such as the origins of embarrassment and shame, the quest after justice even beyond what is justified by deterrence, the inexplicable decline of supply of internal organs once their market is legalized, and so on. Such anomalies would make standard economic theory vulnerable to attacks from other branches of social science armed with non-parsimonious, multiple-self conceptions.
Acknowledgements The paper was supported by research fellowships from the Alexandre von Humboldt Foundation (Germany) and the Earhart Foundation and benefitted from the readings of Amartya Sen, Howard Margolis, Cass Sunstein, anonymous refrees, and especially Steven Joyce. The technical help of Carole Brown is acknowledged. The usual caveat applies. References Albert, Sidney P., 1971. Reflections on Shaw and psychoanalysis. Modern Drama 14 (2), 169–194. Arrow, Kenneth J., 1997. Invaluable goods. Journal of Economic Literature 35 (2), 757–765. Becker, Gary S., 1996. Accounting for Tastes. Harvard University Press, Cambridge, MA. Davidson, D., 1980. How is Weakness of Will Possible? In: Davidson, D., Essays on Actions and Events, ch. 2. Oxford University Press, Oxford. Dworkin, Ronald M., 1977. Is law a System of Rules? In: Dworkin R.M. (Ed.), The Philosophy of Law. Oxford University Press, Oxford. Elster, Jon, 1983. Sour Grapes: Studies in the Subversion of Rationality. Cambridge University Press, Cambridge. Elster, Jon (Ed.), 1986. The Multiple Self. Cambridge University Press, Cambridge. Etzioni, Amitai, 1986. The case for a multiple-preference conception. Economics and Philosophy 2 (2), 159–183. Frankfurt, Harry M., 1971. Freedom of the will and the concept of a person. Philosophical Review 68 (1), 5–20. George, David, 1993. Does the market create preferred preferences?. Review of Social Economy 51 (3), 323–346. George, David, 1998. Coping rationally with unpreferred preferences. Eastern Economic Journaling 24 (2), 181– 194. Hahn, Frank, 1982. One Some Difficulties of the Utilitarian Economist. In: Sen, Amatya, Williams, Bernard (Eds.), Utilitarianism and Beyond. Cambridge University Press, Cambridge, pp. 187–198. Harsanyi, John C., 1955. Cardinal welfare, individualistic ethics, and international comparisons of preference. Journal of Political Economy 63, 309–321. Harsanyi, John C., 1986. Individual Utilities and Utilitarian Ethics. In: Diekmann, Andreas, Mitter, Peter (Eds.), Paradoxical Effects of Social Behavior: Essays in Honor of Anatol Rapoport. Physica-Verlag, Heidelberg, pp. 1–12. Harsanyi, John C., 1986. Rational Behavior and Bargaining Equilibrium in Games and Social Situations. Cambridge University Press, Cambridge. Hausman, Daniel M., McPherson, Michael S., 1993. Taking ethics seriously: Economics and contemporary moral philosophy. Journal of Economic Literature 31 (2), 671–731. Jeffrey, R.C., 1974. Preferences Among Preferences, The Journal of Philosophy 377–391.
386
E.L. Khalil / J. of Economic Behavior & Org. 40 (1999) 373–386
Kavka, Gregory S., 1991. Is individual choice less problematic than collective choice?. Economics and Philosophy 7 (2), 143–165. Khalil, Elias L., 1995. On the scope of economics: What is the question?. Finnish Economic Papers 8 (1), 38–53. Khalil, Elias L., 1996. What is economic action? From Marshall and Robbins to Polanyi and Becker. Journal of the History of Economic Thought 18 (1), 13–36. Khalil, Elias L., 1997a. Buridan’s ass, risk, uncertainty, and self-competition: A theory of entrepreneurship. Kyklos 50 (2), 147–163. Khalil, Elias L., 1997b. Etzioni versus Becker: Do moral sentiments differ from ordinary tastes?. De Economist 145 (4), 491–520. Khalil, Elias L., 1999. Symbolic Products: Prestige, Pride and Identity Goods. Theory and Decision, in press. Kuran, Timur, 1995. Private Truths, Public Lies: The Social Consequences of Preference Falsification. Harvard University Press, Cambrdige, MA. Lewis, Helen Block, 1971. Shame and Guilt in Neurosis. International Universities Press, New York. Margolis, Howard, 1982. Selfishness, Altruism and Rationality: A Theory of Social Choice, Cambridge University Press, Cambridge. McPherson, Michael S., 1980. Liberty and the higher pleasures. In defense of mill. Economic Inquiry 8 (2), 314–320. Mueller, D., 1986. Rational egoism vs. adaptive egoism. Public Choice 51, 3–23. Nethercot, Arthur H., 1969. Bernard Shaw and psychoanalysis. Modern Drama 11 (4), 356–375. Posner, Richard A., 1995. Aging and Old Age. University of Chicago Press, Chicago. Radin, Margaret Jane, 1996. Contested Commodities. Harvard University Press, Cambridge, MA. Rawls, John;, 1971. A Theory of Justice. Harvard University Press, Cambridge, MA. Rawls, John, 1988. The Priority of right and ideas of the good. Philosophy and Public Affairs 17 (4), 251–276. Schelling, Thomas C., 1984b. Choice and Consequence. Harvard University Press, Cambridge, MA. Sen, Amartya K., 1977. Rational fools: A critique of the behavioral foundations of economic theory (Reprinted in Sen, 1982, ch. 4). Philosophy & Public Affairs) 6 (4), 317–344. Sen, Amartya K., 1980-1981. Plural Preference. Proceedings of the Aristotelian Society, pp. 193–215. Sen, Amartya K., 1982. Choice, Welfare and Measurement. Basil Blackwell, Oxford. Sen, Amartya K., 1985. Goals, commitment, and identity. Journal of Law, Economics and Organization 1 (2), 341–355. Sen, Amartya K., 1987. On Ethics and Economics. Basil Blackwell, Oxford. Sen, Amartya K., 1990. Justice: Means versus freedoms. Philosophy and Public Affairs 19 (2), 111–121. Sen, Amartya K., 1993. Internal consistency of choice. Econometrica 61 (3), 495–521. Sen, Amartya K., 1995. Is the Idea of Purely Internal Consistency of Choice Bizarre? In: Altham, J.E.J., Harrison, Ross (Eds.), World, Mind and Ethics: Essays on the Ethical Philosophy of Bernard Williams. Cambridge University Press, Cambridge, pp. 19–31. Sen, Amartya K., 1997. Maximization and the act of choice. Econometrica 65 (4), 745–779. Shaw, Bernard,1931. Three Plays for Puritans: The Devil’s Disciple, Caesar and Cleopatra, and Captain Brassbound’s Conversion. Constable and Company, London, 1900. Stewart, Hamish, 1992. Rationality and the market for human blood. Journal of Economic Behavior and Organization 19 (2), 125–143. Sunstein, Cass R., 1997. Free Markets and Social Justice. Oxford University Press, Oxford. Taylor, Charles, 1995. A most peculiar institution. In: Altham, J.E.J., Harrison, Ross (Eds.), World, Mind and Ethics: Essays on the Ethical Philosophy of Bernard Williams. Cambridge University Press, Cambridge, pp. 132–155.