Groupe de REcherche en Droit, Economie, Gestion UMR CNRS 7321 INTENTIONAL APPLE-CHOICE BEHAVIORS: WHEN AMARTYA SEN MEETS JOHN SEARLE Documents de travail GREDEG GREDEG Working Papers Series Dorian Jullien GREDEG WP No. 2013-46 http://www.gredeg.cnrs.fr/working-papers.html Les opinions exprimées dans la série des Documents de travail GREDEG sont celles des auteurs et ne reflèlent pas nécessairement celles de l’institution. Les documents n’ont pas été soumis à un rapport formel et sont donc inclus dans cette série pour obtenir des commentaires et encourager la discussion. Les droits sur les documents appartiennent aux auteurs. The views expressed in the GREDEG Working Paper Series are those of the author(s) and do not necessarily reflect those of the institution. The Working Papers have not undergone formal review and approval. Such papers are included in this series to elicit feedback and to encourage debate. Copyright belongs to the author(s). Intentional Apple-choice Behaviors: When Amartya Sen Meets John Searle JULLIEN Dorian* GREDEG Working Paper No. 2013-46 Abstract This paper suggests that Amartya Sen’s conception of rationality could benefit from insights borrowed to John Searle’s philosophy of mind. More precisely, I argue that the work of Searle on intentionality provides a relevant conceptual apparatus to strengthen Sen’s conceptualization of context-dependent preferences in a way that suggests further analytical contributions to the latter’s line of research. The arguments developed in the paper are relevant for three interrelated issues on economic rationality that are currently discussed in economic methodology: (1) methodological dualism and intentionalitic explanations in economics, (2) the relationships between economics and philosophy, and (3) the recent rise of behavioral economics within the mainstream of economic theory. Key words: rationality, intentionality, preferences, context-dependency, Amartya Sen, John Searle Résumé. Le papier suggère que certaines dimensions de la conception de la rationalité économique développée par Amartya Sen peuvent se consolider au travers de la philosophie de l’esprit de John Searle. Plus précisément, la théorie searlienne de l’intentionnalité fournie un appareil conceptuel pertinent pour renforcer la conceptualisation senienne des préférences dépendantes au contexte. Les arguments développés dans le papier concernent trois problèmes contemporains qui sont interconnectés autour de la rationalité économique et discutés en méthodologie de l’économie : (1) le dualisme méthodologique et les explications intentionnalistes en économie, (2) les relations entre économie et philosophie, et (3) l’émergence récente de l’économie comportementale dans le ‘mainstream’ de la théorie économique. Mots clefs : rationalité, intentionnalité, préférences, dépendence au contexte, Amartya Sen, John Searle JEL: B00, B40, B41, B49 * GREDEG, Université de Nice Sophia-Antipolis, [email protected]. I thank Richard Arena, Rosa Volpe, Lauren Larrouy, Cyril Hédoin, John Davis, and especially Danielle Zwarthoed for their helpful comments. All potential blunders are of course under my own responsibility. 1 Introduction Amartya Sen has been an important actor of the last four decades in articulating the relationships between economics and philosophy in a constructive manner. One line of research in Sen’s approach to economic rationality consists in conceptual and formal inquiry into the context-dependency of preference relations (Sen 2002a, 2002f); and his work in ethics, moral and political philosophy has explicitly influenced his contributions on context- dependent preferences. In this paper, I suggest that the Senian line of research on the context- dependency of preference relations could benefit from insights borrowed to the philosophy of mind. I argue that John Searle’s philosophy of mind (1983, 1992) provides a relevant conceptual apparatus to strengthen the foundations of Senian context-dependent preferences, and to suggest further analytical contributions in this line of research. More precisely, I will emphasize how the conceptual apparatus developed by Searle in his account of intentionality highlights the ontological dimension of the Senian context- dependency of preferences. Intentionality is a central concept in the philosophy of mind, where it is usually defined as “mind’s directedness at objects and states of affairs, the property of some mental states (thoughts, beliefs, desires) to be about something, their ‘aboutness’” (Rakova 2006, p.90). Sen’s account of context-dependent preferences is made through the use of what he calls “external references to behaviors” (developed in the first section), which, roughly speaking, consists in taking into account behaviorally unobservable features of choices in the formalization of context-dependent preferences. The central argument of this paper is that, with respect to the work of Sen (2002a, 2002f), the Searlean conceptual apparatus for intentionality provides a neat way of conceptualizing the way preferences emerge and guide rational choices through the interactions between individuals’ minds and choice contexts. The development of this argument has to be understood against the background of contemporary issues discussed in economic methodology on the conception of rationality in economics (especially in microeconomics), which I will briefly sketch in this introduction. Sen’s avoidance of the topic of intentionality in his account of context-dependent preferences is mainly due to the fact that the philosophy of mind is usually considered as a separate field of inquiry from ethics, moral and political philosophy. Less separated from ethics, moral and political philosophy is the philosophy of action where discussions around 2 the concept of intention are sometimes related to the topic of intentionality in the philosophy of mind. It is understandable that Sen avoided the highly controversial discussions of intention in the philosophy of action (Setyia 2010) to concentrate on the broader issues discussed in his work in ethics, moral and political philosophy. Nevertheless, it has to be noted that the lack of discussion of intentionality in the work of Sen is representative of the state of affair in economics, as has been pointed out by several contributions in economic methodology1. Some of these contributions have already discussed the potential relevance (and limits) of John Searle’s philosophy of mind for economics or subfields and schools of thought in economics2. Other methodological contributions have already stressed the relevance of Searle’s philosophy for the work of Sen3. Taken together, these contributions tackle three central and interconnected methodological issues concerning the conception of rationality in economics: (1) methodological dualism and intentionalistic explanations in the social sciences (i.e., how the methodology of social sciences ought to depart from the methodology of natural sciences due to the specificity of human actions), (2) the relationships between economics and philosophy, and (3) the recent rise of behavioral economics within the mainstream of economic theory. A very enlightening discussion of these points is made by Matthias Klaes (2008). As he puts it: “Empirical studies may explain choice behavior in various proportions of individual and social framing effects. This by itself will do little however to advance a conceptual understanding of bounded rationality. A convincing theory of bounded rationality will have to confront the issue head on. Methodologically, this requires 1 See Wade Hands (2001, pp.66-69, pp.142-143, and esp. pp.334-341) and John Davis (2003 pp.94-96) for insightful appraisals of these contributions, and of the problem of intentionality in economics, economic methodology and philosophy more generally. 2 See, among others (i.e., excluding papers that focuses exclusively on the relevance of Searle’s philosophy of society for economics), the 2002 special issue of the Journal of Economic Methodology on the philosophy of Searle (Boehm 2002, Runde 2002, Faulkner 2002, Viskovatoff 2002, Boettke and Subrick 2002, Horwitz 2002), some of the papers in the 2003 special issue of the American Journal of Economics and Sociology on the philosophy of Searle (Viskovatoff 2003, Schmid 2003 and the review of the issue by Faulkner 2005), and the (very insightful) paper by Mathias Klaes (2008). 3 Nuno Martins (2009) develops Sen’s implicit social ontology through Searle’s philosophy of society, and Cyril Hédoin (2013) develops Sen’s criticisms of the standard conception of rationality in economics through Searle’s (2001) criticisms of the standard conception of rationality in philosophy and the social sciences. 3 engagement with the thorny issue of human intentionality. […]Yet, given that rational choice theory traditionally conceived does not necessarily purport to be a psychological theory, it seems more than apt to pursue the bounded rationality theme into philosophical debates on practical reasoning.” (Klaes 2008, p.217 and p.219) This paper is structured in two sections, in which the arguments discussed are indirectly relevant to these methodological issues. A first section discusses the role of Sen’s formalization of context-dependent preferences in his broader conception of rationality developed through cross-disciplinary discussions between economics and philosophy. There, I propose an ‘apple-choice behaviors’ thought experiment inspired by Sen to illustrate the conceptual apparatus of his formalization of context-dependent preferences, and I discuss the reasons and drawbacks of the lack of discussions with the philosophy of mind in his framework. A second section illustrates how Searle’s conceptual apparatus for intentionality can strengthen the Senian account of context-dependent preferences; I do so by means of applying Searle’s conceptual apparatus to some parts of the thought experiment of the first section. A conclusion summarizes the main points of the argument, and discusses more directly their relevance for the methodological issues (1), (2) and (3) discussed in this introduction. 1. Comprehensive descriptions and external references to behaviors in Sen’s formalization of context-dependent preferences 1.1 Economics and Philosophy in the work of Sen on rationality The criticisms Sen has addressed to the standard conception of rationality in economics target both the a priori impositions of internal consistency of choice conditions on choice-functions in revealed preference theories and the broadening of self interested motives in preference-based utility functions. Sen has pointed out that both approaches are in fact intertwined around the notion of utility so that the interpretations of choice behaviors in economics amount “to get an empirical rabbit out of a definitional hat” (2002b, p.27). For Sen, the conception of rationality as internal consistency of choice makes no sense without any external reference to behavior because, without such external reference, it implicitly underlies that rational agents only act self-interestedly, which is a psychological motive that 4 the consistency of choice approach sought to eradicate (see esp. Sen, 1973, 1987, 2002b). In normative economics, where rationality is understood in terms of choices that are equated with welfare maximization, the intertwinement of consistency of choice with self-interest poses rather dangerous problems: not only it biases observations of actual behaviors as all being guided solely by the pursuit of individual welfare maximization, but it also biases normative policy recommendations by interpreting welfare criteria such as Pareto optimality as being attainable only through self-interested behavior (see esp. Sen 1987). The burden of responsibility for this state of affair is, Sen argues, carried by both (1) the lack of analytical clarity on the relationships between the concept of preference and the concepts of choice, utility maximization, motive and welfare, and (2) the relationships between what he calls the “direct” and the “indirect” purposes of rational choice theory (RCT) in economics (Sen 1987, pp.11-12). On Sen’s account, the current state of affair is that the direct purpose of RCT is used for characterizing the substance of rationality (i.e., the constraints for a choice to be rational, such as consistency conditions or self-interest maximization): it should help us to “think and act wisely and judiciously, rather than stupidly or impulsively” (Sen 2002b, p.42). The indirect purpose of RCT is to describe and predict actual behavior, by first characterizing the substance of rationality (i.e., the direct purpose), “and then assuming that actual behavior will coincide with rational behavior” (Sen 2007, pp.20-21). He observes that, in economics, “[t]he two uses – directly normative and indirectly predictive – are closely linked. Indeed, the latter is basically parasitic on the former, but not vice versa” (Sen 2002b, p.43). Sen departs from this view by turning it upside down so to speak: he develops the normative direct purpose of RCT from its positive indirect purpose. In other words, he theorizes the substance of rationality with insights obtained from actual behaviors. He does so through interdisciplinary exchanges with ethics, moral and political philosophy, to provide insights on the plurality of motives underlying individual choices and on alternative welfare criteria (see esp. Runciman and Sen 1965; Sen 1974, 1987)4. This research strategy led Sen to 4 Though the theorizing from actual behavior to the substance of rationality is also, to some extent, the research strategy in standard decision theory (on this, see Guala 2000, Starmer 2005, Jallais and Pradier 2005, Mongin 2009), the interdisciplinary exchanges with ethics, moral and political philosophy is proper to Sen and his followers. 5 criticize the traditional positive/normative dichotomy in economics. He argues that an explicit account of the value judgments that are at play in actual individual decision making could not only improve normative economics regarding welfare criteria, but also improve the substance of rationality in positive economics. His position in economics has been conjointly developed with respect to the broader philosophical problem of the fact/value dichotomy by himself and other philosophers (see Putnam 2002, Sen 2005, and the edited volume by Putnam and Walsh 2012). Roughly speaking, Sen takes the following methodological stance in ethics, political and moral philosophy: he retains the need of public reasoning (or public deliberation) emphasized by Rawls and other transcendental approaches to justice, but rejects their methodological use of a priori concepts such as the veil of ignorance, for a more comparative and empirically grounded approach to justice (Sen 1980, 2006, 2009)5. Of importance here is the intrinsically social dimension it gives to his approach: ethical and moral norms of behavior are to be found ‘out there’ so to speak, by scrutinizing the “‘triple entanglement: of fact, convention and value’” (Sen 2005, p.112) in society to provide relevant information for understanding these norms, and, by implication, human rationality6. 1.2 Context-dependency of preferences in the work of Sen Sen’s work in ethics, moral and political philosophy provides him the material for formal and axiomatic contributions to the direct purpose of RCT in economics to fight against “[t]he authoritarianism of some context-independent axioms” (Sen 2002b, p.6) of preferences underlying both the rationalization of choice functions and the enlightening of self-interest in the maximization of utility function. One line of conceptual contribution developed by Sen consists in analytical inquiries into the nature of external references to behaviors that can 5 However, see Ragip Ege, Herrade Igersheim and Charlotte Le Chapelin (2012) for a thorough assessment of Sen’s position between transcendental and comparative theories of Justice. 6 Developing the direct purpose of RCT from its indirect purpose also characterizes the postwar developments of economics-oriented decision theory (see Guala 2010, Starmer 2005, Jallais and Pradier 2005, Mongin 2009), but the development of the direct purpose from its indirect purpose through cross-disciplinary discussions with ethics, moral and political philosophy is proper to Sen’s research strategy in economics. 6 rationalize choice behaviors which contradict either the most basic internal consistency of choice conditions or the broadest ‘enlightenment’ of self-interest maximization7. I would like to emphasize a key methodological element in this line of contributions, namely: the ““comprehensive” description of choice processes and outcomes” (Sen 2002f, p.159, my emphasis) from which Sen derives formal and conceptual contributions to RCT. These contributions start from the premise that preferences underlying choices “over comprehensive outcomes (including the choice process) have to be distinguished from the conditional preferences over culmination outcomes given the acts of choice” (Sen 2002f, p.159), where culmination outcomes basically refer to the standard consequentialist view of the relationships between preferences, choices and outcomes (i.e., either chosen outcomes are identified with preferences or preferred outcomes are identified with choices). Comprehensive outcomes and comprehensive choice processes are descriptions of choice behaviors violating the prima facie reasonableness of either the weakest internal consistency of choice conditions or the broadest ‘enlightened’ self-interest maximization. It is from these descriptions that external references to behaviors that rationalize choice behaviors are extracted and formalized. To illustrate the comprehensive description of choice processes and outcomes, let’s start by considering the following thought experiment (that is inspired by comprehensive descriptions given by Sen 2002a, 2002f, and extended in order to illustrate more of his conceptual apparatus at once in the analysis that follows): Imagine a fictional character named Mike, who is at a dinner guest party where there is a plate that, once, was full of apples. At one point, Mike feels like eating an apple. When Mike reaches for the plate there is only one remaining apple so that he has the choice between ‘taking the last’ apple and ‘taking nothing’. Mike knows for a fact that (a) if there had been two apples he would have picked one, as he in fact did the day before at the same kind of dinner guest party facing the exact same choice situation but with one extra apple in the plate, or that (b) if someone else, say, the host, handled him the last apple, he would have accepted the offer, as it also indeed happened the day before, after Mike ate his first apple. But because (c) there is just one apple left and that nobody handles it to him, he prefers to take nothing. Just as he reached this decision, the drunkest guest of the party sees that Mike is fixing the last apple, understands what is going on in Mike’s head, approaches Mike and (d) jokingly yells across the room “to hell with good manners” while handling the last 7 See Sen 2002a, 2002b, 2002c, 2002e, 2002f and the references therein for the characterizations of these context-independent axioms, basic internal consistency conditions and broadenings of self-interested motives. Sen does not deny the existence of rationally self-interested behaviors in some economic contexts (on this, see Sen 1984). 7 apple to Mike: ashamed, Mike declines the offer and responds by pretending that he was absolutely not thinking about that. In this story, Mike faces four choice situations (a) two apples in a plate, (b) the last apple handled by the host, (c) the last apple in a plate, and (d) the last apple handled by the yelling drunkest guest. If we give up on the comprehensive description, the culmination outcomes collapse into the two choice menus: {‘apple’; ‘apple’; ‘nothing’} for situation a, and {‘apple’; ‘nothing’} for situations b, c and d. According to internal consistency of choice approaches to revealed preference, Mike’s choice behavior is consistent (hence Mike is rational) between situations a and b, and between situations c and d, but Mike’s choice behavior is generally inconsistent (hence Mike is irrational) across the four situations: his revealed preferences violate every weak internal consistency of choice condition referenced by Sen (2002a, 2002f). But all the same, given the comprehensive description Mike cannot be deemed irrational since there are reasons to explain his choice behavior in every situations. The variety and multiplicity of such reasons can be very wide. Sen (2002a, 2002f) considers non-exhaustive lists of non-mutually exclusive reasons, which can be separated into two broad groups. On the one hand, there can be reasons that imply an external reference to behaviors related to the identity of the chooser (Sen 2002f, pp.161-162): (1) “Reputation and indirect effects”: e.g., Mike wants to have an ‘apple lover’ reputation in situation a, a ‘good guest’ reputation in situation b, to avoid an ‘apple grabber’ reputation in situation c, to avoid being associated with the drunkest guest and/or the reputation of being a person that cares too much about other people judgments in situation d (2) “Social commitment and moral imperatives”: e.g., Mike could have an implicit or explicit self-imposed ethics of never taking himself the last apple whatsoever in situation c (3) “Direct welfare effects”: e.g., Mike might genuinely enjoy the eating of an apple in situations a and b, and by sympathy, also genuinely enjoy that someone else will benefit from the last apple in situations a and c (4) “Conventional rule following”: e.g., Mike is following rules of ‘proper dinner guest party behavior’ in situation a, b, c and d 8 On the other hand, there can be reasons that imply an external reference to behaviors related to the nature of the choice menus (Sen 2002a, pp.130-131): (5) “Positional choice”: e.g., the culmination outcome ‘apple’ is comprehensively not the same when there is another ‘apple’ and ‘nothing’ in the choice menu, i.e., it is ‘another apple’ in situation a, as when there is just ‘nothing’ in the choice menu, i.e., it is ‘the last apple’ in situations b, c, and d (6) “Epistemic value of the menu”: where one extra option in a choice menu brings information about the situation, including extra information about the other options of the menu. This type of reason does not fit our story very well; Sen usually illustrate these by the example of a person who “given the choice between having tea at a distant acquaintance’s home (x), and not going there (y) […] chooses to have tea (x), [but] nevertheless choose to go away (y), if offered – by that acquaintance – a choice over having tea (x) going away (y), and having some cocaine (z). The menu offered may provide information about the situation” (Sen 2002a, pp.130-131) (7) “Freedom to reject”: when a choice is made not for the sake of the chosen culmination outcome, but rather for the sake of comprehensively rejecting the other outcomes of the choice menu. This type of reason also does not fit out story very well; Sen (ibid) illustrates this as follow: “The point of fasting in the form of not eating (y), given the possibility of eating well (z), may become less clear when the only alternative is to be partly famished anyway (x)”. According to Sen (2002f, pp.162-163), in economics, rationalization of choice behaviors by (1) are typical of the formalization of self-interest maximization, which can also rationalize by (3) when interdependent preferences are taken into account in utility functions; while existing formalization in evolutionary game theory can rationalize by (4). But none of these approaches can rationalize by (2) or by mixtures of all of these reasons while keeping analytical tractability. It is also important to notice that rationalizations by (5), (6) and (7) of a choice function cannot be made through a redefinition of consistency conditions that would index outcomes on their respective choice sets because that would render these conditions “vacuous” since 9