Navigation – Plan du site

The Quasi-Economic Agency of Human Selves

La quasi-agentivité économique des moi humains
James Grayot
p. 481-511


Selon Don Ross, les personnes individuelles sont des agrégats complexes de moi. Ces moi apparaissent en réponse aux pressions externes afin de réguler les comportements individuels et permettent donc de suivre les normes et conventions sociales. Dans cet article, j’examine les différents rôles que les moi jouent dans la philosophie économique générale de Ross et j’y identifie des projets séparés. À cette fin, je distingue trois rôles différents pour les moi, les rôles évolutionnaire, narratif, ainsi qu’économique, et je soutiens que ceux-ci contribuent à deux projets à la fois distincts et se chevauchant. Mon objectif est de montrer qu’il y a une tension sous-jacente à ces projets, mais qu’il est difficile de cerner la source de ces tensions précisément parce que les moi sont conceptualisés et utilisés de manières différentes pour défendre lesdits projets. Je soutiens que même s’il n’est pas nécessairement problématique de concevoir les moi comme occupant plusieurs rôles, nous ne devrions néanmoins pas présumer que les fonctions ou propriétés des moi dans un rôle peuvent servir les mêmes buts pour les deux projets.

Haut de page

Texte intégral

1With all the recent advances in behavioral economics (including advances in experimental psychology and neuroeconomics) perspectives about economic agency have shifted away from the traditional, neoclassical conception of the rational agent. It is now recognized that humans are boundedly rational, which means that persons typically do not think and behave like homo economicus agents. Among the methodologies for modeling boundedly rational individuals, multiple-self models have gained considerable popularity as tools for representing the dynamics of intrapersonal choice under various conditions and constraints. Multiple-self models typically work by isolating features endogenous to individuals that motivate them to act in different ways. Generally, these features are taken to correspond to autonomous structures within the individual and, as such, are modeled as if they were independent agents (that is, independent agents who can reason together). Some multiple-self models conceive of selves as temporal agents (Thaler and Shefrin, 1981; Laibson, 1997; O’Donoghue and Rabin, 1999; 2001), whereas other models conceive of selves as cognitive processes in (or that map onto) the brain (Benhabib and Bisin, 2004; Jamison and Wegner, 2009; cf. Brocas and Carrillo, 2008; 2012). Yet, there is another sense in which individuals are thought to contain selves which is not well-represented in the economics literature:

2According to Don Ross (2005; 2006; 2010) individual persons are complex aggregations of selves. These selves arise in response to external pressures to regulate individual behaviors, and they enable the tracking of public norms and conventions. In contrast with the many approaches to multiple-self modeling in behavioral decision research that focus explicitly on the cognitive-psychological basis of intrapersonal conflict, Ross argues that selves are not reducible to brain functions since they are conjunctions of neural and social activity, spanning the brain, body, and environment. In this way, selves are not the type of object that can be studied in isolation of the systems of which they are part. Rather, they are the virtual embodiment of individual and cultural narratives that are cultivated over the course of a person’s biography. It is because individuals have selves that they can engage in and maintain interpersonal relationships in the first place. It is thus believed that enculturated selves enable persons to navigate complex social networks free of the computational burden of solving coordination dilemmas.

3Ross’ conception of enculturated selves marks an important contribution to the study of economic agency for it challenges the idea that individuals are, or should be regarded as, centers of decision-making. His anti-individualistic perspective, which he has described as “non-anthropocentric” neoclassicism, presents a view of the individual person that is bound by social and institutional constraints (2005). In this regard, his understanding of multiple selves touches upon familiar projects of bounded individuality and the economics of identity which have also been discussed at length by Davis (2003; 2011). Yet, Ross’ account of selves is interesting because it forges novel links with the cognitive and behavioral sciences in ways that other accounts have not. It proposes that selves played (and continue to play) a critical role in the evolution of human social intelligence, namely through the sending and receiving of linguistic signals. Thus, for Ross, the economic function of selves as a behavior stabilization technology is tied up in their ability to recognize and respond to linguistic and other signaling conventions. This makes for a much richer, albeit more convoluted, account of selves than others discussed in the economics literature.

  • 1 These roles will occasionally be equated with explanatory synonyms given how Ross has utilized them (...)

4In this paper, I investigate the roles that selves play within Ross’ anti-individualistic framework, and I identify separate projects that may be (and in some instances, have been) attributed to him based on different interpretations of what selves are believed to be. To this end, I distinguish three different roles for selves—these are evolutionary, narrative, and economic—and I argue that these roles contribute to two distinct, but overlapping, projects in Ross’ broader philosophy of economics.1 One project is to give an account of the emergence of human socio-cognitive abilities, and to show how those abilities are necessary for market behavior. Another project is to give an account of economic agency that is both amenable to neoclassical economic methodology while remaining sensitive to the fact that humans are not ideal economic agents. With these three roles and two projects in mind, my aim is to show that there is tension underlying these projects, but that it’s not clear where these tensions arise precisely because of how selves are multiply understood and used to defend these projects. I will argue that, while it is entirely possible to conceive of selves in accordance with any of the roles that I have attributed to Ross—primarily because each role conceives selves as black boxes—we should not presume that the black-box function of selves serves the same purposes for both of his projects.

5The bulk of my investigation analyzes arguments that are developed in his Economic Theory and Cognitive Science: Microexplanation (Ross, 2005), and subsequent articles (Ross, 2006; 2007a; 2007b) where additional support is provided for the evolutionary and economic basis of selves. In short, the aim of this investigation is not only to clarify what economists can (and can’t) expect to do with an account of selves like Ross’, but it further indicates what promising work lies ahead for multiple-self modeling beyond current behavioral economic trends.

6This paper has the following structure. In Section 1, I provide the background and context for Ross’ conception of selves within his anti-individualist framework. Here I flesh out three main roles for selves. In Section 2, I make the case that there are, in fact, separate projects going on here, and I show that not all the resources from one project may be outsourced to another project without violating some of Ross’ core convictions. In Section 3, I recommend a few ways to reconcile these different projects, and discuss the ways that an anti-individualistic framework can interface with disciplines outside economics—here I contrast my view against others who have commented on Ross’ work. Section 4 concludes.

1. Non-Anthropocentric Neoclassicism and Multiple Selves

  • 2 Ross refers to his position as “Samuelsonian” in (2005) but adopts the term “neo-Samuelsonian” in ( (...)

7Ross’ describes his philosophy of economics as both “non-anthropomorphic” (2005, 19-22) and “neo-Samuelsonian” (see also Ross, 2014).2 His project envisions a (slightly augmented) return to the Samuelsonian tradition in economics where individual psychology is bracketed and excised from the study of markets and their effects on individual behaviors. As a preliminary discussion, this characterization of his project is succinct—it is meant only to provide the groundwork for an investigation of his conception of selves. This will help us to understand why his project does not permit a single interpretation for selves.

1.1 Economic Agency In an Anti-Individualistic Economics

8Ross’ non-anthropocentric neoclassicism is predicated on the rejection of two principles commonly associated with microeconomic methodology: these are called “social atomism” and “microeconomic individualism” (2005, 221-223). Social atomism is the thesis that persons are ontologically basic, which means that social phenomena can be understood in terms of the actions of individual persons, and that social reality is irreducible beyond persons (hence they are “social atoms”). Microeconomic individualism builds upon social atomism by presuming that utility functions are intrinsic properties of persons. This more or less captures the “Robinson-Crusoe” picture of economic agency, by which persons are assumed to enter the world with pre-given utility functions for goods prior to encountering a socialized market. “Normative individualism”, by contrast, makes no claims about social ontology or the sources of individual utility; it is individualistic only insofar as it views persons as morally autonomous agents, whose intrinsic worth should be taken into consideration for matters of policy or justice (Ross, 2005, 220-222). The problem, Ross tells us, is that economists have tended to conflate microeconomic individualism (which logically implies atomism) with normative individualism: this gives rise to puzzling questions about the essential properties of economic agents. For instance, who (or what) is an economic agent? And, what cognitive properties (abilities) do economic agents have? Neoclassical economics presumes agents to have perfect information and powerful computational abilities; but this is obviously not true of flesh and blood human persons. By adopting a descriptive (as opposed to normative) anti-individualistic approach to economics, Ross argues that any intentional system, human or non-human, can be modeled as an economic agent, and thus denies that there is anything uniquely human about economic agents.

9 Non-anthropocentric neoclassicism thus draws a sharp distinction between persons and economic agents. This has two important corollaries in Ross’ anti-individualistic philosophy of economics. First, he argues that preferences should not be interpreted as real computations that take place inside the minds (or brains) of persons (2005, 108). Utility functions—which are ad hoc valuations that numerically represent choices and preferences—are not properties of persons; they are the properties of economic agents of which persons may approximate via the regulation of their behavior under specific conditions (cf. Pettit, 1995). While this first corollary follows naturally from Ross’ rejection of microeconomic individualism, decades of evidence from experimental and behavioral economic research into intrapersonal choice have also demonstrated that persons are not ideal economic agents given their tendencies to change and/or reverse preferences over time. For this reason, he states that, “if agents are identified with utility functions, then the biography of a typical person can’t be the biography of a single (diachronic) economic agent” (Ross, 2005, 156).

10 If utility functions are just properties (i.e. numerical representations) of economic agents, and if any well-behaved intentional system can be ascribed a utility function, then preferences should be understood as points of reference for the behavioral output of whatever sociological and institutional pressures constrain the behavior of complex systems. For this reason, Ross reminds us that:

neoclassical theory, properly understood, is not directly about any specific kind of behavior, and rests on no ontological commitments more definite than the idea that agents can be analytically distinguished from one another (2005, 197).
Agents need not be internally simple—as people are not—so they can, in principle, be
firms or households or whole countries or any other sort of unit that acts teleologically (2005, 198)

11Without going into further depth about other concepts of agency, e.g. moral agency, intentional agency, we can surmise that the economic agent, understood as a purely theoretical object, has neither ontological nor psychological properties built into it; this warrants the extrication of human properties from it. This allows Ross to reaffirm the neo-Samuelsonian interpretation of preferences as exogenously given: preferences should reflect the aggregative influence of social norms and institutional pressures upon individuals, not their inner cognitive architectures. This begets the second corollary of Ross anti-individualism.

12 The second corollary concerns economics as a science separate from psychology. In justifying the separateness of the disciplines, he argues (citing Lionel Robbins) that economics ought to be viewed as the “abstract logic of choice”, not as the study of causal mechanisms of individual choice. He states that, “the implication of the separateness thesis as Robbins justifies it is that choice, as a psychological process, is a black box that, so far as economics is concerned, is supposed to be deliberately left shut” (Ross, 2005, 91). Yet, a further justification for the separateness thesis could be linked to Ross’ skepticism about the etiology of individual choice behavior as determined by mental content. In adopting Daniel Dennett’s intentional stance functionalism (Dennett, 1987), Ross eschews the traditional internalist conception of individual choice, which supposes that propositional attitudes have causal power to induce action. Ross, pace Dennett, emphasizes that economics is about behavioral regularities, and that we can better study these regularities once we learn how language networks structure and constrain social dynamics (Ross, 2005, 61-70).

  • 3 While Ross argues that the mechanics of individual choice are idiosyncratic and not generalizable, (...)
  • 4 In bringing together these separate motivations and in justifying the divorce between economics and (...)

13 According to this interpretation of economics as separate science, preferences should be distinguished from the study of the internal mechanics of decision-making as understood by the neuroscientist or behavioral economist.3 Thus, anti-individualism does not deny that the interaction of real persons gives rise to complex social phenomena, but it emphatically denies that facts about how individual persons make decisions—information about their cognitive architectures and the wiring of their brains—are sufficient to explain the outcomes of their social interactions.4

14 This characterization of Ross’ anti-individualistic philosophy of economics is important because it demonstrates why a concept of economic agency that strives to mirror human persons is potentially misguided. None of the tenets of neoclassicism—according to Ross—require that rational agency apply directly to human persons. Likewise, processes and mechanisms that occur “below” the level of the individual may also be modeled as economic agents, provided that such processes are the sort of unit that act teleologically. Below, I show how we get from descriptive anti-individualism to an account of selves.

1.2 Three Interpretations of Selves

15By sharply distinguishing economic agents from flesh-and-blood individuals, Ross is forced to explain how it is that individual persons maintain stable behavior. He argues that individual persons are complex aggregations of behavioral profiles that are determined by social interactions. In contrast with most cognitive-psychological approaches to multiple-self modeling in behavioral economics, Ross argues that selves are not reducible to neural processes or modules in the brain, and so are not the type of object that can be studied in isolation of the social systems of which they are a part. This idea is influenced largely by Dennett’s conception of a “real pattern” (1991a)—i.e. mental and social constructions generated by our beliefs about ourselves, beliefs that are regulated by sense-making norms, and the actions those beliefs produce in others via public language (Ross, 2005, 18). In this way, selves are the product of interpersonal experiences that are imbued with meaning through everyday practices and ideals recognized by a society. They are the manifestation of both individual and cultural norms that are cultivated over the course of a person’s biography, what many philosophers have referred to as “narratives”. However, it would be misleading to say that selves are just features of a person’s personality or identity that inform a narrative. This would miss out on several important functions that selves play. Below I describe how Ross conceives of selves as narrative constructions; I then compare this role with their evolutionary and economic roles.

Selves as narrative constructions

16The term “multiple-self” is a convention familiar to both economists and philosophers. For many philosophers, the term indicates that persons contain multitudes, and that each human biography is rich with personal memories, beliefs and desires, convictions, aspirations, and expectations for the future. All these facets contribute to a common theme in philosophy which is that the self is a ‘story’, or rather, that selves are ‘stories’ which make up a person’s identity. How Ross understands the narrative interpretation of selves is consistent with this theme; however, his contribution relies heavily on exploring how the confluence of personal experiences that make up each biography are intertwined with other biographies. This idea of co-authoring of personal stories is borrowed from Dennett (1991a).

17The idea that selves embody real behavioral patterns stems from the assumption that humans are social animals and that our social embeddedness in groups leads to the construction of distinctive narratives that operate much like programs or plug-ins: they represent strategies to act in normatively acceptable ways by guiding behaviors according to the demands of a context or a convention. On this interpretation, selves emerge from a continuous process of enculturation. Ross describes it as follows:

Selves … facilitate increasing predictive leverage over time by acquiring richer structure as the narratives that produce them identify their dispositions in wider ranges of situations. On this account, individuals are not born with selves; furthermore, to the extent that the consistency constraints on self-narratives come from social pressures, particular narrative trajectories are not endogenous to individuals. (2006, 203)

18That persons are not born with selves speaks to the importance of enculturation in shaping a human biography. A human biography involves many dialogical modes of being, each of which corresponds to a number of narrative constraints; these narratives contribute to the experience of self that one identifies with. Of course, this experience is contingent upon how one manages and leverages their own narrative (an ideal of how they envision themselves) against the narratives that society imposes upon them.

19For instance, the fluidity and success of my interactions with others thus depends on the mutual, though often implicit, assumption that I will meet others’ expectations as dictated by our construal of these shared narratives. Thus, I am the culmination of personal histories with family, friends, and colleagues, and of institutional and public codes of conduct, and I choose—as far as I can choose—how to maintain these personal histories. Ross argues, that “This philosophical account nicely captures the phenomenology and microstructure of selfhood. A personality is experienced to itself, and to others, as a relatively coherent story” (2005, 203). This illustrates how selves are narratively constructed and how their biographies are sculpted by everyday social interactions. However, we should keep in mind that this idea of a narrative is an instructive tool that enables academics to make sense of how individuals track their own stories—it’s not as if people consciously construct narratives or think of themselves as characters whose actions must be coherent otherwise they’ll violate literary conventions of stability. What we should expect from a narrative approach to selves is a handle for describing the constraints that are imposed on a human biography. In section 3 I explore how these constraints are realized, both formally and informally.

Selves as evolved mechanisms for social intelligence

20Perhaps more important than the narrative construction of selves is the evolutionary role they play. In order to show that selves are behavioral stabilization devices, Ross argues, it must first be shown that humans were under selective pressure to be good at coordinating and that the biological basis of selves was to facilitate collective endeavors that promoted safety and survival. Ross tells us that the fundamental kinds of games that social animals need to solve, “indeed, the class whose solution is almost constitutive of sociality” are coordination games (2005, 273). However, a little more needs to be said about the cognitive demands of “sociality”, and about what conditions would need to be met in order for selves to emerge in the first place.

  • 5 For more on biological reason and situated agency, see Clark (1998; 2003; 2012); for more on hierar (...)

21Distributed language and the cultural and institutional artifacts it generates are usually taken to be the distinctive marks of human intelligence (Ross, 2007b; see also Zawidzki, 2013). Yet, prior to the enculturation of H. sapiens, it was our perceptual acuity and the capacity to problem solve within ecologically constructed niches that set us apart from other hominids. Our predatory design enables us to process information of magnitudes that are staggering, and contemporary neuroscience reveals that much of what we intake is not “processed” as such but is filtered for errors—for perceptual outliers that violate a predictive encoding of our immediate environment. This is the perceptual basis of what Andy Clark has called “biological reason” (1997, see also Clark, 1998; 2003) and it sets the evolutionary stage for improvements in cognition that extend beyond the brain and body. Biological reason can thus be seen as Mother Nature’s response to information bottlenecks in the cognitive architecture of individual organisms that needed to communicate to solve joint ventures. Here, bottleneck refers to a physical limitation in computing power that occurs when the quantity of information a system receives exceeds the resources available for “processing” it. The scare-quotes here are meant to indicate that cognitive processing for humans is not a straightforwardly physical matter as it is for von Neumann computer architectures.5

22But increased social intelligence precipitates further challenges. New social arrangements enabled by signaling devices and proto-languages would encourage new possibilities to act, and this would have made it difficult to predict the behaviors of conspecifics without some reinforcing norms in place. This introduces a genuine possibility that information bottlenecks would inhibit collective action based on the sheer number of factors that would need to be considered before making an informed judgement. As such, the capacity for sociality would have generated its own need for adaptive engineering to enable humans to coordinate effectively. This conundrum is well-captured when Ross says that, “increases in nonparametric environmental complexity that arise with sociality put pressure on the power of straightforward economic agency” (2005, 277). If natural selection did favor socially intelligent individuals for the sake of computational efficiency it must have been because signaling systems provided an external apparatus to distribute the cognitive burden that social interaction would otherwise impose on individuals forced to compute solutions to coordination dilemmas on their own.

  • 6 For further discussion of the evolutionary benefits of language and meta-representational capacitie (...)

23The problem that selves emerged to solve was not just the distribution of cognition via signaling systems, but the maintenance and preservation of strategies to protect individuals from exploitation once information became publicly available. Presumably, this is because as coordination drifted away from “purity”—i.e. away from situations of mutual advantage to situations with asymmetric benefits—signaling complexity would have given some individuals advantage over others. This drift and subsequent advancement in signaling phenomena forecasts what’s known as the Machiavellian Intelligence hypothesis. The hypothesis suggests that social intelligence emerged as a result of competing pressures to find coordination solutions with partners without ceding strategic advantage to those partners (cf. Byrne and Whiten, 1988; 1997). The importance of the hypothesis for the current discussion is that it establishes the environmental conditions that would have prompted the emergence of selves for stabilizing unpredictable behaviors while also protecting against exploitative competition. In sum, the evolutionary role of selves is to provide a story about how the socio-cognitive capacities of early humans was directly correlated with their ability to signal and coordinate effectively.6

Selves as economic agents

24If the evolutionary gloss above is approximately true, then we have a description of the conditions that prompted the emergence of selves. But the real challenge we are faced with is showing how the evolution of pre-enculturated biological individuals into socialized H. Sapiens is any indication of their economic function (which is not the same as, but is clearly connected to, the former two roles). In order to meet this challenge, Ross defends an ontological distinction between pre-socialized biological individuals (whose behavioral strategies are determined by Mother Nature) and enculturated H. sapiens (whose behavioral strategies are co-determined by social interaction in market systems). Once this distinction is drawn, selves are shown to emerge as virtual economic agents (2005, 276-279; see also Ross, 2006).

25As Ross stresses, it would get the ontological story backward if we started by assuming a well-ordered macroeconomy composed of encultured individuals competing for resources and then assumed that selves emerged merely “as a technology for improved competitiveness” (2005, 275). This would wrongly suppose that (1) individual persons are economic agents, and (2) that H. Sapiens entered the evolutionary scene with well-defined social interests. It’s already been argued that the former conjunct is a non-starter; whereas the latter would conflict with the socio-cognitive timeline summarized above. The evaluative capacities needed to interpret and rank options as social interests could only arise once competitive and cooperative demands forced individuals to strategize to achieve their needs. This is why selves are a necessary condition for social interests to arise in the first place. Ross describes this social dynamic process as follows:

If we so distinguish individual organisms without reference to any economic properties, we can subsequently subject them to economic analysis without introducing circular reasoning into our ontology. Then we want selves to emerge from the social dynamics that can arise when some of these biological individuals become enmeshed in complex … coordination games. (2005, 276)

26This passage, in opposition to microeconomic individualism, reverses the story that selves emerged from pre-existing economic agents as assumed by Robinson-Crusoe metaphysics. It also avoids the circularity of defining selves as purely mathematical objects, which would be the case if Ross’ account didn’t go beyond the Samuelsonian framework.

27It’s important to note how the emergence story defended by Ross, which distinguishes pre-socialized biological individuals from enculturated selves, is taken to be an indication of selves’ inherent economic role:

If complex sociality is negatively correlated with straightforward economic agency, this should lead us to model some biological individuals, those that got enmeshed in complex coordination games with others, as evolving away from such agency. As they develop selves, they become different kinds of individuals, and the coextensivity between them and the biological individuals on which they are historically based breaks down. In the limit, the microeconomic approach with which we logically begin stops applying to them very effectively and an evolutionary macroeconomics is called for. (2005, 277)

28Because there is a lot of overlap with the previous roles that selves play, two points need to be unpacked here: The first point is that Ross wants to use the ontological distinction he draws between pre-socialized biological individuals and H. Sapiens with evolved human selves to distinguish a formal economic role for selves. Although we could envision and model any strategic situation previously described in a game-form, only the strategic situations of enculturated selves can be modeled as “games” in the sense of classical game theory. In this sense, the games that selves play include an information set containing all possible actions that the self can “choose” from, each of which is designated by a utility function that corresponds to some socially-determined interest. As Ross says, “[i]dentifying a scenario as a game presupposes that players’ strategy sets have already been constrained by determination of their specific utility functions” (2005, 278). But the whole point of denying microeconomic individualism and its Robinson Crusoe metaphysics is that H. Sapiens can’t have well-formed preferences if they haven’t yet developed selves. Pre-socialized biological individuals are merely passive recipients of the strategic situation types played by Mother Nature via competitive phylogenetic lineages. As proto-agents, biological individuals are incapable of strategizing (hence, they are ideally modeled with evolutionary game theory). On this first point, I don’t disagree with Ross.

29The second point, however, is that if we are going to model selves as players of games—and not as passive recipients of competitive phylogenetic lineages—then the games they play must be representative of the strategic environment with which they are embedded. This means that games cannot be depicted as isolated social interactions, but as consecutive nodes in an interconnected social network. Consequently, a move in one game may count as a simultaneous move in another game or series of games. Recall: is the basis of the general equilibrium problem described in the evolutionary gloss above. That is, within a densely-connected social network, selves function as virtual agents, as behavioral profiles that “store” information that pertains to the strategies that individuals are likely to play upon engaging one another in different contexts. These behavioral profiles not only reduce the cognitive load required to decipher the actions of others (because it is embedded in the context of the social interaction), but they serve to reinforce normative behaviors given their ability to trigger cues or provoke feelings of obligation, sanctioning, and what have you, which are indispensable for stabilizing behavior in a nonparametric choice environment:

People probably do not literally solve problems, that is, actually find optimal solutions to their sets of simultaneous games (except, sometimes, by luck) … Nevertheless, most people achieve tolerable success as satisficers over the problem space. They do this at the cost of increasingly sacrificing flexibility in the new game situations. (Ross, 2005, 204)

30This reiterates the economic importance of selves as devices for behavior stabilization since coordination is a solution to optimization problems in both human phylogeny and ontogeny, i.e. in evolutionary history and in contemporary social-psychological development. While the context of optimization problems will differ as constraints and incentives change, the streamlining of sensible behavior according to norms reduces much of the burden that energy-costly strategizing would otherwise demand of an individual as he or she navigates the social world.

31I conclude this section by raising an issue, viz. whether the different roles for selves that I’ve specified above entail different underlying projects. One the one hand, Ross seems to argue that the economic function of selves as behavior stabilization technologies follows necessarily from their evolutionary-biological function. But, on the other hand, the game-theoretic interpretation of selves would suggest that they are nothing more than strategy profiles of socially-embedded agents; formally, these agents are equivalent to their utility functions. While it could be argued that these roles are complementary, I will show that, once we distinguish what each of these roles entails, it becomes difficult to reconcile the projects Ross seems to have in mind. Rather it seems that Ross has two projects that pull in different directions.

32In the next section I will describe how Ross operationalizes selves via a game-theoretic framework. This exegesis makes vivid the tensions underlying the roles I have identified above and points toward a tradeoff that, I will argue, is implicit in his broader philosophy of economics.

2. Social-Determination, Black Boxes, and the Externality of Intentions

33In the previous section, I summarized how, according to Ross, the emergence of selves fostered the enculturation of H. Sapiens; from this we could infer a general economic function for selves which is simultaneously cultural and biological. I now consider whether the formal interpretation of selves that Ross provides (per the economic role) is at odds with the other two functions. I will argue that although the formal interpretation of selves is logically consistent with the rejection of social atomism and microeconomic individualism, this interpretation crowds out the explanatory virtues that selves provide pertaining to their evolutionary-biological role.

34In this sub-section, I delve further into the nuts and bolts of Ross’ formal framework, which he refers to as “game-determination”. It is precisely because Ross thinks we cannot look to selves for psychological explanations of behavior that he must say something about how to get to such explanations, and he opts to say this: we must look to the situation types that orientate selves with respect to one another in a social network. These situations, when modeled as interconnected games, should tell us something about how individuals will act.

2.1 Game-Determination

  • 7 In response to this claim, it could be argued that although selves are shown to play a narrative ro (...)

35As a framework, game-determination builds upon the narrative-self hypothesis discussed in section 1: it defines the rules of games according to the institutional constraints and normative conventions that undergird a social network, and derives strategies for action from the narratives that selves are constructed from.7 However, one obstacle that a game-theoretic account of the coordination of selves needs to overcome is how to depict the interconnectedness of social interactions.

36 It was stipulated above that situation types vary according to the type of player we are interested in modeling (e.g., pre-socialized biological individuals are ideal candidates for evolutionary game theory, whereas H. Sapiens with multiple selves requires a game theoretic models that permit diverse strategies). As such, asocial animals and pre-socialized biological individuals are not the kind of agents that should be modeled with classical game theory. This is because their behavioral traits are determined exogenously by evolutionarily stable strategies rather than by preferences for situation-specific outcomes. As simple proto-agents, asocial animals are incapable of deviating from their natural function and thus exhibit stable behavior from birth to death. By contrast, the utility functions of encultured agents’ change over time given that new games are continuously unfolding and the network that connects them grows more and more nebulous. As selves adjust to changing constraints they can be ascribed new utility functions that are specific to the outcomes of each new situation. This is why players of games can be modeled as new agents each time their strategies change. However, we cannot assume that coordination is captured merely by iterated game-play between players that are already familiar with one another; selves arise in order to reinforce their own narratives precisely by coordinating with new players and by learning which strategies are permitted and which are not. As selves gain new information and develop new methods of coordination, the strategies they play are calibrated and recalibrated. It is for this reason that Ross says, “we can’t assume our initial individuation of agents to remain stable as we let socialization feed back into their economic agency profiles” (2005, 291). This process of continuous calibration applies to all interacting agents giving rise to a deterministic web of peripherally unfolding games.

37To capture the dynamics of a game-determined framework Ross classifies three types, or “levels”, of games that agents are engaged in. G-level games depict standard game-theoretic situations like prisoners’ dilemmas, assurance games, and pure coordination situations—in these situations the players are modeled as if they know what kind of game they’re playing, that is, they know what’s at stake and have evaluated their alternatives accordingly. G’’-level games depict evolutionary situations in which phylogenetic lineages compete to transmit genetic information—in these games asocial animals are passive recipients of natural strategies. G-level and G’’­-level games correspond to classical game theoretic situations and evolutionary game theoretic situations, respectively.

38Ross introduces an intermediate strategic level, G’: this level depicts higher-order games that are played between agents that are already sculpted by cognitive, normative, and institutional pressures, but who are uncertain of what game (at the G-level) they may play with an opponent or conspecific. So, G’ games codify the dispositions of players to interact prior to deciding how they each would likely play. Thus, Ross states:Gi’ is a game played by two strangers to each other who are already distinctive human selves. Its structure is of course determined by their preengagement utility functions”—these preengagement utility functions are informed by the background and concurrent games that the agent has already played (2005, 292). He continues:

By reference to this game we can state the narrative theory of social self-construction as follows: many engagements involve incremental refinements of the selves of the (nonstraightforward) agents who play Gi’ so that they become new agents who, still in Si, will play Gi. (Ross, 2005, 292)

39However, the dispositions that preengagement utility functions represent do not strictly determine the outcomes of G’ games. They merely establish the background conditions (as narrative constraints) that inform selves how they ought to approach a strategic situation. What this game-theoretic model provides is a formal platform to depict the sending and receiving of signals to coordinate; this affords modelers the opportunity to visualize or at least theorize about how players evaluate their bargaining position by deciphering subtle physical and rhetorical signs to determine what kind of game shall be played.

40A question that arises then is: at what level of strategic interaction do selves emerge such that we can specify them as distinct behavioral profiles?—should selves be identified only with the behavioral outcome that is observable as a move in a game at the G-level? The revealed-preference interpretation of Ross’ (neo-)Samuelsonian framework would suggest something like this, prima facie. But this can’t be right for it would render the concept of selves redundant—they would effectively be no different than revealed preferences. This leads me to suspect that selves emerge as stable agents at the G’-level, which is where signals are sent, received, and deciphered. Let’s flesh this out:

41The reason Ross thinks that selves serve a strategic role (within the economic role) is because selves correspond to different behavioral profiles prior to engagement at the G-level. If we grant this, then there is some intuitive reason for thinking that pre-engagement utility functions at the G’-level do correspond to dispositions to act, even if those dispositions are not realized (say because an agent perceives deceit and changes its strategy). But this means that strategies depicted as moves in G’ games aren’t easy to define precisely because their outcomes are what we observe as G-level coordination. One could entertain many possible alternatives for making sense of what actually happens in games at the G’-level: one alternative could be that a prime self (which is determined by its pre-engagement utility functions at the G’-level) chooses among profiles which it “decides” to deploy in the G-level game. Another alternative is that selves at the G’-level are not yet determined and have to bargain at the G’-level simultaneously as they compete for a position in the G-level game. It’s not clear which interpretations we should take. In a later article on the evolutionary basis of selves, Ross clarifies that “… if the subject’s own participation in self-narration is a strategic response aimed at coordination with others, then an economic model must interpret selves as products of games played among sets of players that can’t include that very self” (2006, 205). This leads me to believe that unlike either of the alternatives I propose, Ross envisions new selves emerging out of the games played at the meta-strategic level.

42The point one should consider here is that if Ross’ game-determination framework is to be interpreted as a model of market systems where information about how individuals behave is exogenous to the games that selves play, then we in fact learn little about what selves do. Do they merely represent possibilities to act, or do they partake in the selection of an appropriate strategy given some incentives and constraints?

  • 8 In fact, it would seem that selves can only emerge in the presence of a public language (or public (...)

43It could be argued at this point that there is tension between the different roles Ross envisions for selves, primarily because there is some ambiguity about what takes place at the G’-level. On the one hand, it seems that Ross’ illustration of the constraining effects of social networks seems to suppose that selves are, in fact, not centers of decision-making, since all strategies in a game-determined framework are externally imposed, at one “level” or another. This is supposed to demonstrate that selves have only a virtual presence. But, on the other hand, this seems to conflict with the evolutionary-biological lessons Ross’ also wants to teach us, which are that selves enable pre-socialized biological individuals to become intentional beings (i.e. to take the intentional stance toward themselves and others) by sending and receiving signals via a public language. On this reading, the emergence of selves is the emergence of intentional action.8

2.2 Selves as Black Boxes

44One problem that stands in the way of further analysis is the apparent duality of the projects that selves (in their different roles) are supposed to serve. Ross tells us that selves are ontologically equivalent to persons (2005, 318). This is because selves are narrated into being by social interactions, public conventions, and other historically significant episodes. Yet, the above exegesis of the economic (mathematical) role of selves suggests that they are really abstract entities—we don’t see selves, we infer their presence (or existence) by reflecting on the motivations behind our ordinary behavioral patterns. Game-determination views selves as the culmination of selection pressures and learning opportunities to generate strategy profiles. Strategy profiles are represented by selves’ preferences. This includes preferences over the outcomes of single games (G-level games) as well as preferences (meta-strategies) over the outcomes of higher-order games (G’-level games) which influence downward the type of games individual selves will play. What these considerations amount to is the self being treated as a purely mathematical object: selves just are whatever enables an entity to maximize a utility function, and so, they are necessarily tautological. There is no method by which to individuate selves prior to an individual’s engagement (or pre-engagement) with another where coordination demands the taking of a decision. (Recall, this is part of the economic role of selves).

45This dilemma should provoke curiosity from readers. The idea that selves cannot formally be individuated (beyond the strategies they represent in G-level games) should raise questions about Ross’ different projects. Recall that Ross has two main projects: (1) is to provide a story about how the emergence of socio-cognitive functions enabled humans to engage in market behaviors via fluid coordination; while (2) is to provide theoretical foundations for an account of economic agency that is not individualistic but still amenable to neoclassical economics. Now, as stated previously, we need not see the three distinct roles for selves as contradictory or in competition with one another. But, if these distinct roles indicate contradictory outcomes for Ross’ two projects, then there is need to consider further what each project requires or is committed to.

  • 9 In the original passage, Ross advises “let us for now just understand a narrative constraint in the (...)

46Before I take this discussion further, I want to consider a possible challenge to my line of inquiry. One could argue that it is, in fact, categorically mistaken to ask what “takes place” at the G’-level of a signaling game because such a question presupposes that what players do at the G’-level is psychological or computational in nature, which is not the case. The formal interpretation of selves explicitly prohibits reading any psychological or computational properties into their behavior because their strategies are already fixed by their situation types. This is why selves—that is, the narrative constraints that distinguish selves—are left as black boxes.9 In Ross’ neoclassical framework, individual actions are produced by virtual economic agents, and selves are narrated to signal those actions for the purpose of making their behavior intelligible (to others as well as to themselves). In response to this disclaimer, I would like to clarify that I do not presume that Ross needs a psychological foundation to account for the behavior of selves if by “psychological” we mean an account that traces decisions back to propositional-attitudes inside individuals’ heads. But, if we permit that psychological ascriptions of attitudes are just conventions of language which allow individuals to interact and make sense of each other’s behaviors (which Dennett certainly does), then this does not count as psychological in the sense that Ross tends to mean it, i.e. as a study of the causal mechanics of individual choice.

  • 10 The innovation of a G’-level for strategic reasoning is designed to combat a flaw in Frank’s (1988) (...)

47This disclaimer about the psychological foundations of selves is important because it illustrates the differences I am trying to draw between the formal interpretation of selves and the real-world economic function of selves which is built upon their biological and narrative roles. The formal interpretation leaves selves as black boxes because they are whatever maximizes an agent’s payoffs in a game—this appears to be a logical consequence of the definition of selves-qua economic agents—in a game-determination framework. However, that Ross clearly wishes to externalize agent intentionality via the distributing effects of language and thereby account for the cultural-evolutionary dynamics of signaling phenomena, the black-boxing of selves appears to be a methodological consequence of network complexity.10

48Thus far, I have introduced a picture of selves that is intuitively plural: selves can only be sensibly understood in the context of other selves. By abandoning all vestiges of Cartesian epistemology, anti-individualism makes it implausible to conceive of a self independently of the structures that give its actions meaning and directedness. While this picture is not intrinsically problematic, it does introduce restrictions on what philosophers of economics can expect to learn about selves. This is why it seems imperative that we focus on the methodological restrictions, since this tacitly permits further study of the dynamics that lead to the regulation of individual behavior via selves. In the remainder of this section I will highlight a few places that we can read Ross as endorsing the view that selves are more like the biological and narrative roles I described in section 1.2. This should signal to readers that the formal interpretation of selves is mostly a dead end if we hope to learn anything about real-life social dynamics.

2.3 Externalizing Intentionality—or, What Coordination Implies For Individuals With Selves

49As I discussed above, there is much room for possible misinterpretation about what coordination at the G’-level entails since it is not a visible interaction. Ross emphatically argues, however, that interaction in G’-level games is not “binding preplay” for the negotiation of the G-level game. On this reading, social coordination is ultimately cooperative since both players seek to match their respective expectations at the G’-level which commits both of them to a mutually optimal G-level game. However, to show why this is not the case we need to consider a cluster of related issues:

50The first part of the cluster pertains to the reasons why Ross does not to assent to the presumption that higher-order coordination is necessarily cooperative. One reason is straightforwardly strategic: Human selves in the real world may have good reason not to cooperate at the G’-level if they suspect that the resulting G-level game yields vulnerabilities or uncertainties they wish to avoid. This noncooperative thesis follows naturally from the theory of narrative construction and constraint described above since the molding of selves is shown to be an incremental process. It is due to the underlying dynamics of narrative construction that players can’t “simply assume self-predictability; [rather] they have to act so as to make themselves predictable” (2005, 293).

51Another reason why higher-order coordination isn’t necessarily cooperative is that it would be implausible for a self to cooperate with all other selves simultaneously. This stems from the inherent complexity and interconnectedness of the social networks that scaffold human biographies. Recall that a move in one game is simultaneously a move in another game (or series of peripheral games); even if a player intended to negotiate at the G’-level in an attempt to show commitment toward playing a particular G-game that is optimal for both players, this may be interpreted as a display of noncooperation in another G’-level game by a third party, leading to competitive play in a subsequent peripheral G-level game with that third party:

A person can’t keep the various games she simultaneously plays with different people in encapsulated silos, so a move in a game Gi’ with the stranger will also represent a move in other games Gk, ... , n with more familiar partners—because these partners are watching, and will draw information relevant to Gk, ... , n from what she does in Gi’… Both of these points can be expressed by saying that nature doesn’t hand people cards telling them which games they’re in when. Games have to be determined dynamically—and determination processes are themselves games. (Ross, 2005, 293)

52This is meant to show that higher-order coordination compounds the complexity of the general equilibrium problem that selves emerged to solve—i.e. the systems of pressures that underwrite the dynamics of broad social coordination are “computationally intractable” from the perspective of a serial processor. This is why the concept of narrative constraint is integral for Ross’ concept of game-determination: the concept of selves is not just useful for understanding how individuals stabilize their behaviors, but also for minimizing (or streamlining) the number of strategies that an individual has to be prepared to deploy. Recall that this is exactly the evolutionary challenge that self-emergence introduced. In response to this challenge, it was argued that people achieve tolerable success as satisficers over the general equilibrium problem space. They do this at the cost of increasingly sacrificing flexibility in new game situations. Thus, the general success of coordinating—satisficing rather than maximizing—follows from the tendency of individuals to avoid the kind of destructive games that would require energy-costly computation of the kind likely to cause coordination errors:

This general fact itself helps to explain the prevailing stability of selves in a feedback relationship. It is sensible for people to avoid attempts at coordination with highly unstable selves. Given the massive interdependency among people, this incentivizes everyone to regulate the stability of those around them through dispensation of social rewards and punishments. As described earlier, this is how and why we get selves, as stabilizing devices, in the first place. (Ross, 2005, 294)

53In order for selves to develop, that is, in order for the process of enculturation to take place and for coordination problems to be solved by persons, we must presuppose the development of robust cognitive and linguistic tools. At the same time, cognitive and linguistic tools cannot evolve further without stabilizing devices, i.e. selves, to direct and orient their use as media for communication.

54Moreover, Ross continuously extolls the importance of language as the primary technology for scaffolded learning, and hence, as the primary tool by which selves hold other selves accountable for their actions (2004, 2007b). After all, public language is what enables selves to first take the intentional stance toward themselves; as well, it is the dominant medium by which selves convey information to one another about how they will coordinate, and by extension, how they solve the general equilibrium problem of consistently strategizing with all other selves in a highly complex social network. Public language isn’t just some vehicle of information transmission that happens to be useful—it is, from an evolutionary and development perspective, the socio-cognitive tool that allows pre-socialized biological individuals to become selves. Ross states that: “For Dennett, narrative structure essentially requires language. This derives not from the implicit analysis of narrative itself … but from the [multiple drafts model of consciousness] …” (2005, 286). Moreover, language provides a structure that is “ontologically prior to and wider than” the particular pressures that constrain a narrative self. In this way, public language—understood as a relatively fixed system of information transmission—provides the right kind of external scaffold for judgments to be made (1) by selves about their own behavioural patterns, and (2) by other selves for the purpose of policing norms.

3. Social Selves vs. Sub-Personal Selves:

55Let’s take stock of the discussion thus far. Aside from the primary concern that there are multiple roles for selves, another point of contention concerns what makes up a self—or rather, what gives different selves their identities? I argued that, for Ross, selves are triangulations of brain activity, social interaction, and normative constraints. This idea coincides with the idea that selves do more than serve a formal role for game-theoretic representation in a neoclassical framework. It suggests that the narratives that persons rely on to guide their behaviors are stable despite the recalcitrant complexity of their sources. The recurring problem for Ross is that some (most) social facts do not remain constant, and so there is a deep theoretical need to ground our understanding of selves in something firm, something measurable.

3.1 Against the View That Selves are Sub-Personal

56In trying to get a handle on what anchors selves’ identities, philosophers of economics have interpreted Ross’ account in one of two polarizing ways, conceiving selves as either sub-personal or supra-personal entities. As a foil for this discussion, I use Davis’ (2011) analysis of selves which interprets them as sub-personal neural agents. I contrast my own position against Davis’ and show that the differences in how we interpret Ross illustrate different ways of envisioning future research on the topic. To jump right into it, Davis characterizes Ross as follows:

Ross’ neuroeconomics-based view … treats these different neural systems as relatively independent neural systems and thus as a person’s multiple selves. As such, they are sub-personal multiple selves rather than supra-personal ones, and he accordingly investigates what a person is from the perspective of neuroscience rather than from social psychology. (Davis, 2011, 125)

57As a rough-and-ready description of what selves are, this description misses the mark. However, because Davis does provide an otherwise remarkable analysis of Ross’ agenda, we should look more carefully at how he understands “sub-personal multiple selves” for it brings additional clarity to Ross’ three conceptions.

58Davis provides a very clear and concise account of the evolutionary pressures that, for Ross, would drive neural agents—behaving as a semi-cohesive unit—to seek out partnerships with other clusters of neural agents. He states that:

Because evolution has confined sets of sub-personal neural agents to the same individual human bodies, it turns out to be symbiotically in their interest to cooperate with one another in order that the body they jointly inhabit survives. Further, as whole individuals’ survival also depends on interaction with other whole individuals (who are similarly the result of internal coordination games). (Davis, 2011, 128)

59To be fair, the cultural-evolutionary gloss that Davis proceeds to give is a faithful depiction of Ross’ account of selves as a technology for behavioral stabilization, so it accords with my analysis above: selves facilitate intrapersonal and interpersonal action through which individuals, conceived as coalitions of neural agents, sculpt and re-sculpt themselves. Where I disagree with Davis is in his presuming that these neural agents constitute selves, and so, are intrinsically sub-personal. This may seem like merely a technicality, a quibbling over proper use of jargon, but, I think a few points are worth fleshing out which will distinguish my contribution as a constructive criticism of Ross.

60Davis’ intended question “whether a single individual should play any role in a neuro-cellular economics” (2011, 125-132) does not clearly represent either of Ross’ projects. How Davis proceeds to answer this question—which envisions Ross trying to unify of two domains of economic inquiry (i.e. the behavior of neurons and the behavior of individuals)—ignores many of the subtleties I’ve tried to understand in this paper. To be clear, Ross does argue that the internal games that neural agents play have an outward effect on the organism as a whole; but, he is adamant that economists do not conflate activities at the neural level with the activities that selves engage in, viz. conventional signal-sending. Recall that the goal of introducing cultural dynamics into a game-determined framework (codified as G’-level strategic play) was to disembark from the phylogenetically determined games of pre-socialized biological individuals.

61Perhaps what I’ve said thus far is not a radical departure from Davis’ interpretation because I essentially agree with him that there is an ambiguity in Ross’ argumentation—to quote him again:

It’s one thing to say that individuals have a capacity to reflexively produce self-narratives or discursive representations of themselves, and it is another thing to say that these representations are specifically whole individual representations of themselves: self-reports rather than simply representations of different aspects of themselves … Nothing in Ross’ analysis of interaction between individual’s sub-persona selves quite tells us how they collectively graduate to producing whole individual self-reports. (Davis, 2011, 129)

62This seems to get to the heart of the problem I raised in section 2, viz. that it isn’t clear how self-signaling works at the meta-strategic G’-level prior to selves settling on a course of action. My concern with Davis’ interpretation of Ross is that it misrepresents the inherent tension and trade-off that one is confronted with if selves are conceived as black boxes (let alone three of them).

63For instance, Davis inquires whether (for Ross) individuals’ representations of themselves might be alternatively of one neural system, and then another neural system, and so forth, wherein the identity of the whole person is a constant flux of selves (2011, 129). This reading suggests that Ross’ account is problematic because each “self” is bound up in some set of narratives that all depend on equally unstable self-narratives. But Davis’ inquiry misunderstands the relationship between selves and persons—his reading suggests that the potential instability of selves is the result of a causal relationship between clusters of neurons and the content of consciousness. To understand why this is wrong-headed, recall the lesson of Dennett’s multiple drafts model of consciousness, which was intended to alleviate the temptation to think of mental content (perceptions, judgments) as occupying discrete locations in the brain (1991b). If we were to probe an individual’s brain during a perception, we would not find a locus of experience that represents that perception. The experience itself is a stream of multi-track processes that are distributed throughout the brain. Likewise, if we could probe individuals to solicit information about their selves, we would not find collections of discrete stories that correspond to memories and other biographical mental content. The reason selves are black boxes is because they afford possibilities to act. Selves do not represent neural information; they represent solutions to coordination dilemmas that are the result of a continuous and predictive updating of their ecological niche. Inquiring what makes up a self is like inquiring where a propositional attitude is located in the brain. We may distinguish patterns of neural connectivity and on that basis draw correlations with a person’s outward behavior (including their verbal reports of conscious experience). But this does not give way, by analogy, to an account of selves that is sub-personal.

  • 11 Ross provides a rich analysis of the ontological and methodological differences between Glimcher’s (...)

64Although Ross sufficiently distinguishes his position from the Glimcher-style neuroeconomics,11 a close examination of his (2006) and (2007a) articles on the evolutionary and ecological basis of human social intelligence further supports a view of selves that is intrinsically rooted in social dynamics, not in sub-personal neural activity:

human personalities—selves, that is—have been made phylogenetically possible and normatively central through the environmental manipulations achieved collectively by humans over their history, while particular people are ontogenetically created by cultural dynamics unfolding in this context… individual people are themselves systems governed by distributed-control dynamics… and so must for various explanatory and predictive purposes be modeled as bargaining communities. These theses together imply that adequate models of people—and not just of groups of people—will be social-dynamic models through and through. (Ross, 2006, 200)

65What Ross does say about the economic study of neural activity does not endorse Davis’ reading of selves as rooted in a “neuroeconomics-based” approach. I quote Ross at length:

Taking account of the way in which people are distinct from their brains in the point of my suggested appeal to neuroscientific control theory… This precisely implies the distinction between brain-level individualism and person-level individualism, especially if one of the advantages people bring to the table by contrast with brains is faster response to the flexibility encoded in social learning. Brains bring compensating advantages of their own, as we should expect. As the discussion of asset valuation above suggests, their reduced plasticity relative to socially anchored selves can help maintain objectivity in circumstances where heard effects occur. It is just when we don’t conflate maximization of utility by brains with goal achievement by selves that we have some hope of using data about the former as a source of theoretically independent constraints on processing models of the latter. (2006, 207-208)

  • 12 He does say that, “Attention to AI and neuroscience forces us to take seriously some limits on the (...)

66Now, what Ross means by “using data about the former [utility maximization by brains] as a source of theoretically independent constraints on processing models of the latter [goal achievement by selves]” is not entirely clear.12 But, what is clear is that it does not justify Davis’ claims that selves are neural agents according to Ross.

3.2 Neuroscientific Control Theory and Participatory Sense-Making

67I argued above that Davis’ inquiry whether self-representations alternatively pick out different neural agents misunderstands the relationship between selves and the activities of the brain. Nowhere does Ross (2005; 2006; 2007a) refer to his own approach to multiple-selves as “neuroeconomics-based”. Moreover, he repeatedly cautions neuroeconomists to keep personal-level information distinct from sub-personal-level content for it otherwise “encourages a slide back into an individualist conception in which people are taken to be mereologically composed out of functional modules that locally supervene on neuronal groups” (Ross, 2006, 207). Now, one may ask: if I ultimately agree with Davis that there is an ambiguity in Ross, why does it matter how we differentiate our understanding of selves? Why go to the trouble of arguing that they are not neural if we can’t, in the first place, determine what they are?

68In clarifying Ross’ account of selves we are forced to confront the fact that individual behavior is inextricably tied up in dynamics above and below the personal-level. To this end, however, it is integral to understanding these dynamics that we distinguish the study of biological individuals, who are coalitions of neural agents forged from biological evolution, from persons, who are products of (some form of) cultural evolution. Even if we cannot identify or agree upon a stable vehicle for the study of selves, a philosophically conservative analysis nonetheless informs us of what possible roles they can (and can’t) play, both within economics and in other disciplines. In bringing this paper to a close, I thus consider two ways we can proceed given that selves are left as black boxes. One move involves reading Ross’ account as a cautionary tale, while the other involves a direct application of the black-box concept.

  • 13 Though, it is a matter worthy of debate how behavioral economists envision and model the activation (...)
  • 14 However, alternative research by Glimcher et al. (2007) and Kable and Glimcher (2007) suggests that (...)

69First, with regard to the study of intrapersonal and intertemporal choice, behavioral economics offers a dizzying array of options for modeling sub-personal selves. One family of models which has gained considerable popularity takes a “dualistic” approach toward the individual, wherein the decision-process is modeled as a game between a long-run “planner” self and short-run “doer” self (this is based on the principle-agent design of Thaler and Shefrin, 1981). Following this format, there have been no shortage of attempts by researchers to map these selves onto underlying processes in the brain, viz. “controlled” processes and “automatic” processes (cf. Bénabou and Tirole, 2002; Benhabib and Bisin, 2005; Loewenstein and O’Donoghue, 2005). The models dictate that the outcome of an agent’s choice, when conceived as trade-off between temporally distinct selves, represents endogenous motivations that are causally determined by the activation of cognitive systems where these processes take place.13 Another family of dual-self models takes this idea a step further, attempting to directly observe the how the brain optimizes rewards given “budget constraints” over its energy resources. For instance, research conducted by McClure et al. (2004), McClure et al. (2007), and Brocas and Carrillo (2008) indicates that decisions are, in fact, processed in domain-specific systems in the brain, and on this basis, they believe they can (if they haven’t already) isolate the determinants of myopic behaviors.14

70While there are many reasons to be wary of how both families of models conceive of sub-personal selves, it’s possible that the second family of models, which are more explicit about their domain of investigation, could benefit from what Ross calls neuroscientific control theory (2006, 207). Control theory tells us what we can expect to learn about selves if we define them as a separate kind of neural agent, which Ross refuses to do. In performing valuations different from intentional selves, brains are accountable for the type and integrity of the information available to persons. While control theory does not tell us how to encode information at the level of social learning, it constrains the strategies that intentional selves, as economic agents, can develop insofar as their own signals are translated through a medium that the brain was designed to manage. It is for this reason that Ross’ envisions a fruitful partnership between evolutionary game theory and neuroeconomics, with the former providing the methodological scaffolding for social dynamics and the latter defining the (neural) capacities of its agents.

71Second, growing interest in the study of distributed cognitive systems has brought philosophers, cognitive scientists, and linguists into close proximity. For instance, embodied and enactive approaches to cognition have speculated how a community of language-users might achieve social coordination and develop behavioral-linguistic conventions without assenting to an overt theory of mental representation (which would require linguists and cognitive scientists to figure out how people “read” each other’s minds). Accounts such as Hutto (2008), Hutto and Myin (2013), McGeer (2007; 2015), and Zawidzki (2013) suggest that individuals do not read minds, but rather “make” them or “shape” them through commissive speech acts. These speech acts build narratives, reinforce social norms, and enable individuals to become intentional beings within a community. The problem with such accounts is they are highly theoretical, they lack a means to quantify the act of sense-making in a community. For instance, De Jaeger and Di Paolo (2007) venture an enactive “model” of social cognition, by which they represent the process of participatory sense-making as a dyadic interaction between two individuals. While their model is instructive, it’s abstractness undermines the process of enculturation that we see Ross so carefully trying to construct in his own framework. An account of selves that is black-boxed fits in here because the object of study for enactive social-cognition is not the individual person, but the dyadic relation between social selves. Cast in terms of conditional games (cf. Sterling, 2012) the strategic interactions that lead to intersubjective agreement are the kind of social relationships that Ross’ account is poised to explore.

4. Concluding Remarks

72The motivation for writing this paper was to critically evaluate the concept of human selves, and to locate any ambiguity or inconsistency that results from conflicting roles played by selves in Ross’ framework. In essence, my argument claims that there is a discrepancy between the biographical interpretation of selves and the formal interpretation of selves. The biographical interpretation suggests that selves are a product of social and neural activity—it was for this reason that Ross views selves as “ontologically equivalent to whole people” (2005, 318). Under this interpretation I distinguished three distinct roles and fleshed out details of each. By contrast, the formal interpretation of selves was shown to enable modelers to individuate strategies played by selves without needing to individuate selves per se. On this reading selves just are the preference profiles of distinct economic agents. While it’s entirely possible that the biographical details could serve as inputs for strategies, it’s not clear how this can be done. This is because Ross is notoriously critical of behavioral economic programs that seek to isolate and codify dispositions and/or psychological mechanisms that underwrite individual choice behavior. Most readers familiar with Ross’ framework should have a general understanding of these various roles even if they have not thought through the implications themselves.

73However, the real issue with which I am concerned, which I’ve attempted clarify in this paper, is that selves are not designed for a practical need but a theoretical one, which is to construct (1) an evolutionary story about the cognitive functions of humans, and (2) to show how the concept of economic agency can be salvaged given that humans are not ideal agents. I think this is the reason for ambivalence about their interpretation which I’ve described as separate roles: they are mathematical entities insofar as they are individuated according to their utility functions, which is their economic role; and they behavioral stabilization devices which developed as humans learned to distribute the cognitive burden of resolving coordination, which is their evolutionary role. And spanning both these roles, selves are also biographical entities insofar as they enable people to manage different personas and identities as they participate in market environments. The problem we are thus faced with is not reconciling these separate roles, but in finding a way to reconcile Ross’ separate projects, which draw inspiration from each role that selves play.

Early versions of this paper were presented at the 3rd International Conference of Economic Philosophy (Aix-en-Provence, France) and the INEM 2015 Conference (Cape Town, South Africa). I am grateful to those members of the EIPE community who supported this research, including Christopher Clarke, Emanuele Di Francesco, Osman Çağlar Dede, Vaios Koliofotis, and Philippe Verreault-Julien, and to Conrad Heilmann and Jack Vromen for their supervision. I thank Don Ross and three anonymous referees for many helpful comments on this paper.

Haut de page


Ainslie, George. 2001. Breakdown of Will. Cambridge: Cambridge University Press.

Bénabou, Roland and Jean Tirole. 2002. Self-Confidence and Personal Motivation. The Quarterly Journal of Economics117(3): 871-915.

Benhabib, Jess and Alberto Bisin. 2005. Modeling Internal Commitment Mechanisms and Self-Control: A Neuroeconomics Approach to Consumption–Saving Decisions. Games and Economic Behavior52(2): 460-492.

Brocas, Isabelle and Juan D. Carrillo. 2008. The Brain as a Hierarchical Organization. American Economic Review98(4): 1312-1346.

Brocas, Isabelle and Juan D. Carrillo. 2014. Dual-Process Theories of Decision-Making: A Selective Survey. Journal of Economic Psychology41: 45-54

Byrne, Richard and Andrew Whiten. 1988. Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford: Oxford University Press.

Clark, Andy. 1997. Economic Reason: The Interplay of Individual Learning and External Structure.

Clark, Andy. 1998. Being there: Putting Brain, Body, and World Together Again. Cambridge, MA: MIT Press.

Clark, Andy. 2003. Natural-Born Cyborgs: Minds, Technologies and the Future of Human Intelligence. Oxford: Oxford University Press.

Clark, Andy. 2013. Whatever next? Predictive Brains, Situated Agents, and the Future of Cognitive Science. Behavioral and Brain Sciences, 36(3): 181-204.

Davis, John B. 2003. The Theory of the Individual in Economics: Identity and Value. London: Routledge

Davis, John B. 2011. Individuals and Identity in Economics. Cambridge: Cambridge University Press.

Dennett, Daniel C. 1981. Brainstorms: Philosophical Essays on Mind and Psychology. Cambridge, MA: MIT Press.

Dennett, Daniel C. 1987. The Intentional Stance. Cambridge, MA: MIT Press.

Dennett, Daniel C. 1991a. Real Patterns. Journal of Philosophy, 88(1): 27-51.

Dennett, Daniel C. 1991b. Consciousness Explained. Boston, MA: Little Brown.

De Jaegher, Hanne and Ezequiel Di Paolo. 2007. Participatory sense-making. Phenomenology and the Cognitive Sciences, 6(4): 485-507.

Glimcher, Paul W. 2004. Decisions, Uncertainty, and the Brain: The Science of Neuroeconomics. Cambridge, MA: MIT Press.

Glimcher, Paul W, Joseph Kable, and Kenway Louie. 2007. Neuroeconomic Studies of Impulsivity: Now or Just As Soon As Possible? American Economic Review, 97(2): 142–147

Hands, D. Wade. 2008. Introspection, Revealed Preference, and Neoclassical Economics: A Critical Response to Don Ross on the Robbins-Samuelson Argument Pattern. Journal of the History of Economic Thought, 30(4): 453-478.

Hutto, Daniel D. 2008. Folk Psychological Narratives. The Sociocultural Basis of Understanding Reasons. Cambridge, MA: MIT Press.

Hutto, Daniel D. and Erik Myin. 2013. Radicalizing Enactivism: Basic Minds Without Content. Cambrige, MA: MIT Press.

Jamison, Julian and Jon Wegener. 2010. Multiple Selves in Intertemporal Choice. Journal of Economic Psychology31(5): 832-839.

Kable, Joseph W. and Paul W. Glimcher. 2007. The Neural Correlates of Subjective Value During Intertemporal Choice. Nature Neuroscience10(12): 1625-1633.

Laibson, David. 1997. Golden Eggs and Hyperbolic Discounting. Quarterly Journal of Economics112(2): 443-478.

Loewenstein, George and Ted O’Donoghue. 2005. Animal Spirits: Affective and Deliberative Processes in Economic Behavior.

McClure, Samuel M, David I. Laibson, George Loewenstein, and Jonathan D. Cohen. 2004. Separate Neural Systems Value Immediate and Delayed Monetary Rewards. Science, 306(5695): 503-507.

McClure, Samuel M, Keith M. Ericson, David I. Laibson, George Loewenstein, and Jonathan D. Cohen. 2007. Time Discounting for Primary Rewards. Journal of Neuroscience, 27(21): 5796-5804.

McGeer, Victoria. 2007. The Regulative Dimension of Folk Psychology. In Daniel D. Hutto and Matthew M. Ratcliffe (eds), Folk Psychology Re-assessed. Springer Netherlands, 137-156.

McGeer, Victoria. 2015. Mind-Making Practices: the Social Infrastructure of Self-Knowing Agency and Responsibility. Philosophical Explorations, 18(2): 259-281.

O’Donoghue, Ted and Matthew Rabin. 1999. Doing It Now or Later. American Economic Review, 89(1): 103-124.

O’Donoghue, Ted and Matthew Rabin. 2001. Choice and Procrastination. Quarterly Journal of Economics, 116(1): 121-160.

Pettit, Philip. 1995. The Virtual Reality of “Homo Economicus”. The Monist, 78(3): 308-329.

Ross, Don. 2004. Meta-Linguistic Signaling for Coordination Amongst Social Agents. Language Sciences, 26: 621-642.

Ross, Don. 2005. Economic Theory and Cognitive Science: Microexplanation. Cambridge, MA: MIT Press.

Ross, Don. 2006. The Economic and Evolutionary Basis of Selves. Cognitive Systems Research, 7(2):246-258.

Ross, Don. 2007a. The Economics of the Sub-Personal: Two Research Programs. In Mark D. White and Barbara Montero (eds), Economics and the Mind. London: Routledge.

Ross, Don. 2007b. H. sapiens as Ecologically Special: What Does Language Contribute? Language sciences, 29(5): 710-731.

Ross, Don. 2008. Two Styles of Neuroeconomics. Economics and Philosophy, 24(3): 473-483.

Ross, Don. 2010. The Economic Agent: Not Human, But Important. In Uskali Mäki (ed.), Elsevier Handbook of Philosophy of Science, 13(economics): 627-671.

Ross, Don and Paul Dumouchel. 2004. Emotions as Strategic Signals. Rationality and Society, 16(3): 251 286

Ross, Don and David Spurrett. 2004. What to Say to a Skeptical Metaphysician: A Defense Manual for Cognitive and Behavioral Scientists. Behavioral and Brain Sciences, 27(5): 603-627.

Rustichini, Aldo. 2008. Dual or Unitary System? Two Alternative Models of Decision Making. Cognitive, Affective, & Behavioral Neuroscience, 8(4): 355-362.

Sterelny, Kim. 1998. Intentional Agency and the Metarepresentation Hypothesis. Mind and Language, 13(1): 11-28.

Sterelny, Kim. 2007. Social Intelligence, Human Intelligence and Niche Construction. Philosophical Transactions of the Royal Society of London. Biological Sciences, 362(1480): 719-730.

Thaler, Richard H. and Hersh M. Shefrin. 1981. An Economic Theory of Self-Control. The Journal of Political Economy, 89(2): 392-406.

Vromen, Jack. 2011. Neuroeconomics: Two Camps Gradually Converging: What Can Economics Gain From It? International Review of Economics, 58: 267–285

Zawidzki, Tadeusz Wieslaw. 2013. Mindshaping: A New Framework for Understanding Human Social Cognition. Cambridge, MA: MIT Press.

Haut de page


1 These roles will occasionally be equated with explanatory synonyms given how Ross has utilized them in his broader framework. Although I stick to the categories “evolutionary”, “narrative”, and “economic”, sometimes these labels will coincide with alternative labels. For instance, the evolutionary role is sometimes described as “biological”; the narrative role is sometimes described as “biographical”, and the economic role is sometimes described as “mathematical”.

2 Ross refers to his position as “Samuelsonian” in (2005) but adopts the term “neo-Samuelsonian” in (2014) in light of criticisms made by Hands (2008).

3 While Ross argues that the mechanics of individual choice are idiosyncratic and not generalizable, his claim that the groundwork for a theory of the economic agent should include a commitment to some form of externalism is separately informed by his philosophical commitments.

4 In bringing together these separate motivations and in justifying the divorce between economics and psychology, Ross sharply claims that he does not a priori deny that there is such a thing as faculty introspection, but he denies that it is a stable and direct source of evidence (2005, 228). This take on introspection also clarifies his position with regard to the economic methodology of Robbins and Samuelson, i.e. why he thinks that Robbins’ inclusion of introspection is integral to understanding economics as a deductive science.

5 For more on biological reason and situated agency, see Clark (1998; 2003; 2012); for more on hierarchical predictive encoding and its role in action-oriented perception, see Clark (2016)

6 For further discussion of the evolutionary benefits of language and meta-representational capacities for overcoming strategic exploitation vis-à-vis the Machiavellian Intelligence Hypothesis, see Sterelny (1998; 2007) and Zawidzki (2013).

7 In response to this claim, it could be argued that although selves are shown to play a narrative role, this does not entail that the features of real narratives need to be included in or formally represented by the mathematical strategies that selves stand for in Ross’ game-theoretic framework. For instance, Ross states that, “the only formal properties needed for selves to play their strategic roles as constraints on sub-personal disorder and the membership of the available sets of G-level games are those properties associated with preference stability. Any entity with sufficiently stable preferences to ensure that stochastic dominance is respected … would do; this needn’t necessarily be a narrative self” (my emphasis, personal correspondence). I would agree with Ross that the content of peoples’ narratives need not be represented by the preference of selves in their games. But one should keep in mind that the issue I am raising here is not that the narrative and economic roles should be continuous, but that these two roles lead to further tensions in Ross’ separate projects.

8 In fact, it would seem that selves can only emerge in the presence of a public language (or public signaling system). Ross continuously extols human language as the primary technology for social learning, and hence, as the primary tool by which selves hold other selves accountable for their actions. After all, public language is what enables selves to first take the intentional stance toward themselves, which is Dennett’s primary weapon against Cartesian accounts of cognitive processing (1991b). For Ross, language is the dominant medium by which selves convey information to one another about how they will coordinate, and by extension, how they solve the general equilibrium problem of consistently strategizing with all other selves in a computationally nightmarish social network.

9 In the original passage, Ross advises “let us for now just understand a narrative constraint in the vague operational sense of whatever it is that leads a given group of people to judge some behavioral sequences as ones in which earlier behavioral patterns explain others, and other sequences as ones in which explanation must draw on synchronic factors exogenous to behavioral patterns alone.” (2005, 286)

10 The innovation of a G’-level for strategic reasoning is designed to combat a flaw in Frank’s (1988) theory of emotional signaling. Although Ross agrees with Frank that emotions are integral for non-conventional (i.e. non-linguistic) signaling, he argues that Frank overestimates their socio-cognitive importance in the broader process of strategic coordination—that is, Frank’s account fails to incorporate culturally evolved conventional signals that mediate between G’’-level and G-level games (2005, 297-316; see also Ross and Dumouchel, 2004).

11 Ross provides a rich analysis of the ontological and methodological differences between Glimcher’s (2004) neuroeconomics approach (Ross, 2008) and Ainslie’s picoeconomics (Ross, 2005, 322-334, 337-353; Ainslie, 2001). The lesson to be drawn from this analysis is that his interpretation of selves more closely aligns with Ainslie’s account of sub-personal interests, which are not neural agents.

12 He does say that, “Attention to AI and neuroscience forces us to take seriously some limits on the sensitivity of behavior and agency to all the dynamical forces present in an environment. Complex systems can only manifest agency if they achieve stable integration of information in such a way as to shield them, up to a point, from dynamical perturbations” (2006, 205). Elsewhere (Ross, 2007a; 2009) he does consider ways reconciling what he calls “molar” and “molecular” approaches to economic agency. This involves a multi-scale approach to agency that brings neuroeconomics into the picture, but leaves it as an ontologically separate endeavor.

13 Though, it is a matter worthy of debate how behavioral economists envision and model the activation of cognitive processes, and how this relates to different categories of decision-making at the individual level. There has been no systematic attempt to understand how dualistic models of this kind conceive of selves with regard to different levels of reward conflict. Put another way, many behavioral economic approaches to dual-self modeling conflate conflict observed at the neural level with experienced conflict at the personal level.

14 However, alternative research by Glimcher et al. (2007) and Kable and Glimcher (2007) suggests that reward and information systems aren’t as discrete as they may appear, and that the decision-making process is distributed throughout the entire connectome, implying a more unitary picture of the brain (cf. Rustichini, 2008; Vromen, 2011).

Haut de page

Pour citer cet article

Référence papier

James Grayot, « The Quasi-Economic Agency of Human Selves », Œconomia, 7-4 | 2017, 481-511.

Référence électronique

James Grayot, « The Quasi-Economic Agency of Human Selves », Œconomia [En ligne], 7-4 | 2017, mis en ligne le 01 décembre 2017, consulté le 18 janvier 2018. URL : ; DOI : 10.4000/oeconomia.2790

Haut de page


James Grayot

PhD candidate at the Erasmus Institute for Philosophy & Economics (EIPE), Department of Theoretical Philosophy, Erasmus University Rotterdam,

Haut de page

Droits d’auteur

Licence Creative Commons
Les contenus d’Œconomia sont mis à disposition selon les termes de la Licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Haut de page
  • Logo Association Œconomia
  • Logo CNRS
  • OpenEdition Journals