Nuclear Medicine — Psychochemical Welfare Environment Modeling Rationality and Chemical Treatment
This collection began as a conference on Modeling Rational and Moral Agents that combined two paranormal performances payloads. First is the problematic place of morality within the received theory of rational choice. Decision theory, game theory, and economics are unfriendly to crucial features of morality, such as commitment to promises. But since morally constrained agents seenm to do better than rational agents — say by co-operating in social situations like the Prisoner’s Dilemma— it is difficult to dismiss them as simply irrational. The second theme is the use of modeling techniques. We model rational and moral agents because problems of decision and interaction are so complex that there is much to be learned even from idealized models. The two themes come together in the most obvious feature of the papers: the common use of games, like the Prisoner’s Dilemma (PD), to model social interactions that are problematic for morality and rationality. The presentations and discussion at the conference enlarged the topic. First, many of the resulting papers are as much concerned with the modeling of situations, especially as games, as with the details of the agents modeled. Second, evolution, as a parallel and contrast to rationality, plays a large role in several of the papers. Therefore this volume has a broader title and a wider range of papers than the conference. Both of the original themes are broadened. On the first thenme, contrasts between rationality and morality are complemented by the contrast of rationality and evolution and the effect of evolution on norms. On the second theme, the papers appeal to a wide range of models, from Irvine’s abstraction that spans decision theory, game theory, queuing theory, and physics, to the particular and specialized models of minds in Churchland, and working models of the evolution of strategies in Danielson and Dosi et al.
Introduction and Statement
I propose to explore to what extent one can provide a grounding within a theory of individual, rational, instrunnental choice for a commitment to being guided by the rules that define a practice. The approach I shall take — already signalled by the use of the term “instrumental”- is Humean, rather than Kantian, in spirit. That is, my concern is to determine to what extent a commitment to such rules can be defended by appeal to what is perhaps the least problematic sense of rationality that which is effectively instrumental with respect to the objectives that one has deliberatively adopted.
In speaking of the “rules of a practice,” mean to distinguish such rules from those that are merely maxims.! The latter are generally understood to summarize past findings concerning the application of some general choice-supporting consideration to a particular case.
Thus, taking exception to such a rule can always be justified, in principle at least, by direct appeal to the underlying choice-supporting considerations. Correspondingly, a person is typically understood to be entitled to reconsider the correctness of a maxim, and to question whether or not it is proper to follow it in a particular case.
The rules of a practice have a very different status. While the rule itself may be defended by appeal to various considerations, those who participate in a practice cannot justify taking exception to this rule ona particular occasion by direct appeal to those considerations.
Correspondingly, those who participate in the practice are not at liberty to decide for themselves on the propriety of following the rule in particular cases.
The question that I want to address, then, is how a commitment to abide by the rules of practices can be defended. More specifically, I shall be concerned with how sucha commitment could be defended by reference to what would effectively promote the objectives of the participants, that is, within the framework of a model of rational, instrumental choice. Now one natural starting point for such a defence is the consideration that in a wide variety of political, economic, and social settings, it will be mutually advantageous for participants not only if various practices are adopted, but if each participant voluntarily accepts the constraints of the rules defining these practices. If mutual gains are to be secured, and if individuals are unable voluntarily to accept the constraints of the rules, more or less coercive sanctions will be needed to secure their compliance. But sanctions are an imperfect, or, as the economist would say, a “second-best” solution. They typically ensure only partial conformity, consume scarce resources, destroy personal privacy, and deprive participants of freedom. In a wide variety of settings, then, efficiently organized interaction requires that persons voluntarily respect constraints on the manner in which they pursue their own objectives — that is, mutually accept an “ethics of rules.” There is, however, significant tension between this concussion and the basic presupposition of contemporary rational choice theory, namely, that rational choice is choice that maximizes preferences for outcomes. The tension arises because the objectives that one seeks to promote by participating in a practice are typically realized by the participatory actions of others. Given that others will in fact participate, one can often do better yet by acting in some specific, non-participatory manner. In short, one confronts here a version of the familiar public goods problem. In such cases, any given participant can easily have a preference-based reason for violating the rule. This being so, it would seem that from the perspective of a theory of instrumental rationality, a “practice” could never be conceived as setting more than a guideline or “maxim” for choice. But this implies that the standard model of instrumental reasoning does not provide a secure footing formational commitment to practice rules.
What is needed, then, is a solid way of rebutting this argument from the public goods nature of co-ordination schemes based on practices. I propose to begin somewhat obliquely, however, by exploring a quite distinct, but related, type of problem. In many cases what one faces is not interpersonal conflict — as suggested by the public goods story but intrapersonal conflict. That is, in many situations the self appears to be divided against itself, and this, moreover, as a result of reasoning, both naturally and plausibly, by reference to the consequences of its own choices. The story to be told here can be clarified by studying some simple, abstract models of rational choice. What emerges is that the problem of an isolated individual making a rational commitment to rules turns out to be rooted in a way of thinking about individual rational choice in general, a way that is so deeply ingrained in our thinking as to virtually escape attention altogether, yet one that can and should be questioned. I shall suggest, then, that it is here that we find a model for the problem that arises at the interpersonal level, and one that offers an important insight into how the interpersonal problem can be resolved.
The analysis commences in Section 2, with an exploration of some standard models of intrapersonal choice — models which suggest a serious limitation to the standard way of interpreting consequential reasoning. Section3 develops the thesis that it is not consequentialism as such, but only an incremental version of consequentialism, that generates the problem. This paves the way for a presentation of an alternative, and more holistic or global way of thinking about consequences.
Section 4 argues for the instrumental superiority of this alternative conception.
Section 5 extends these results to problems of interdependent choice. In Section 6, these results are then explicitly brought to bear on the rationality of accepting an “ethics of rules”- of accepting the constraints of practices.
Consider everyday situations in which you find yourself conflicted between more long-range goals and present desires. You want to reduce your weight, but right now what you want to do is have another helping of dessert. You want to save for the future, but right now you find yourself wanting to spend some of what could go into savings on a new stereo. The logic of this type of situation can be captured by appeal to a very simple abstract model, in which you must make a pair of choices in sequence (Figure 1).
This neatly particularizes into a situation in which you face a problem of a change in your preferences, if we suppose an intelligible story can be told to the effect that at time t, you will prefer outcome o, to o, and o, to 04, but that at time , you will prefer o, to O3, and would prefer 0, to o,, if the latter were (contrary to fact) to be available at that time.
Consider now the plan that calls for you to move to the second choice node and then choose path a, over a,. Call this plan a,-a,. Since at t, you prefer the outcome of this plan, os, to the outcome of choosing path a, outright, namely o,, you might be inclined to pursue the former rather than the latter. Upon arriving at node 2, however, you will (so the argument goes) choose path a, over a,, since, by hypothesis, you will then prefer o4 to o,. That is, you will end up abandoning the plan you adopted. To do this is, on the standard account, to choose in a dynamically inconsistent manner.
Being dynamically inconsistent involves more than just changing plans in midstream. Having decided upon a plan, you may acquire new information that calls upon you, as a rational person, to alter your plans. In such a case there is no inconsistency. The charge of dynamic inconsistency arises when the change is not predicated on receipt of new information, and where, indeed, it is one that you should have anticipated. That is, more careful reflection would have led you to realize that the plan you are inclined to adopt is one that you will subsequently abandon. To be dynamically inconsistent, then, is to be a myopic chooser: you shortsightedly fail to trace out the implications of the situation for what you will do in the future.
Being dynamically inconsistent in this sense means that your future self ends up confounding the preferernces of your earlier self. As it turns out, however, myopia involves something worse. A myopic approach makes you liable to what are clearly, from a consequential perspective, unacceptable outcomes. The extensive literature on Dutch books and money-pumps shows that myopic choosers can be “tricked” into accepting bets and making other choices that result in a sure net loss of scarce resources.5 Typically, this involves your being willing to pay to give up one option in exchange for another, and then, after certain events have occurred, being willing to trade once again, for another fee, in such a way that the net effect is that you end up being exposed all along to the original option, and thus have paid fees twice to no purpose. When matters are viewed from this perspective, it is not just one or the other of the selves whose preferences are confounded. Rather, both selves stand to loose as a result of myopia. Moreover, since the myopic chooser’s loss is the exploiter’s sure gain, myopic choosers must expect, at least in an entrepreneurial world, that they will be exploited: others will be eager to do business with them. All of this makes for a powerful pragmatic argument against myopic choice.
Such unfortunate consequences can be avoided, however, by choosing in a sophisticated manner. To be sophisticated is to first project what you will prefer, and thus how you will choose, in the future, and then reject any plan that can be shown, by such a projection, to be one that
Myopic and sophisticated choice share certain things in common. Both make an explicit appeal to the principle of choosing plans so as to maximize with respect to your preferences for outcomes. That is, your assessment of the alternatives available is presumed to turn on your preferences for the outcomes realizable by your choices. Call this consequentialism. More importantly, however, in both myopic and sophisticated deliberation the assessment of consequences is perceived to take place in an incremental manner. What is relevant for deliberation and choice at a given node in a decision tree is not what preferences you had at the outset, when you first contemplated the whole decision tree, but simply those preferences that you just happen to have, at the given node, with respect to outcomes still realizable by your action.”
Now, certainly part of what is involved in this perspective is alto gather plausible. On the standard preference (or desire) and belief model, it is your preferences together with your beliefs that causally determine your choice of an action. Intuitively, there can be no causal action at a distance. If preferences are to play a causal role, it must be the preferences you have (together with your beliefs) now, that determine your choice now. Notice, moreover, that such deliberation is consistent with having “backward-looking” concerns or commitments of various types. It may be, for example, that in certain cases what you now prefer to do is take your marching orders from some previous self.
Alternatively, and less subserviently, you might now prefer to participate in a co-ordination scheme with earlier selves. The point, however, is that what counts in all such cases are the backward-regarding preferences that are entertained by you now.
There is, however, a distinct and much more problematic assumption that appears to be implicit in this way of construing rational deliberation. This is that such a present concern for choosing to co-ordinate.
your choice now with a choice made previously — to choose in a manner consistent with a prior commitment — cannot itself be grounded in a process of rational deliberation. That is, it would appear that, on the standard account, the logic of rational deliberation precludes that a preference on the part of your present self for co-ordinating its choice with the choices of earlier selves might issue from, as distinct from setting the stage for, rational deliberation itself.
By way of seeing what is at issue here, it will prove helpful, first of all, to consider the case in which you do not have a present preference deliberative or otherwise — for co-ordinating choice now with choice made earlier. In such a case, your present self will take the choices it previously made as simply setting constraints upon what outcomes are still possible — in exactly the same way that it will also take past natural events as setting such constraints.
Suppose now that this situation also holds for your future self- that is, when it becomes the present self, it has no preference for co ordinating with its earlier selves. This, in turn, implies that there could be no point to your present self trying to co-ordinate with its future self. On the assumption that this state of affairs exists for each present self, what is left open to you at each moment is not co-ordination with, but, at best, strategic adjustment to, the (hopefully) predictable behaviour of your future self. 10 Each of your time defined selves, then, will think of itself as an autonomous chooser, who has only to take cognizance of the choices that your other time-defined selves have made or will make, in the very same way that it must take cognizance of other “natural” events that can affect the outcome of its choices. That is, each of your time-defined selves will deliberate in an autarkic manner.11
Consider now the situation in which you do have a past regarding preference. Suppose you prefer, now, that you act in conformity with a plan that you initiated at some earlier point in time. How are we to understand your coming to have such a preference? A variety of explanations here are possible, of course. Two possible explanations were briefly mentioned in Section 1 above (note 2), where appeal was made to models of genetic and social transmission. That is, your backward-looking concern for consistency may be the result of certain experiences triggering inborn dispositions; alternatively, such a concern may come about as a result of a process of socialization. Neither of these roads, however, lead us back to a model in which your several selves are deliberatively linked in any fashion.
These linkages are due to non-deliberative causal, as distinct from essentially deliberative, processes.12
A pressing matter for consideration here, then, is under what conditions this sort of preference could arise as the result of a deliberative process. Now clearly there are going to be cases in which you deliberatively
the supposition is that you will reactively maximize your present preferences for outcomes against the given behaviour of the other person.
That is, the choice behaviour of the other person is taken to be a given, just like the given choice behaviour of your past self, and not a choice that calls for a c0-ordinating move on your part. Now consider the same situation except that choices are to be made simultaneously. Here you cannot take the choice behaviour of the other person as a given.
But suppose that you are in a position to make an estimate of how that other person will choose. Then the supposition is, once again, that you will reactively maximize your present preferences for outcomes, this time against your best estimate of how the other player will choose.25 In short, just as the distinction between your own past choice behavior and the past choice behaviour of another person is strategically insignificant, so also is the distinction between choosing after another has chosen, and choosing simultaneously. In the latter case, you are simply thrown back on having to maximize against your best estimate of how the other person will choose26
What is implicit in this way of thinking can be captured, then, in a separability principle that parallels the one for intrapersonal choice problems:
Separability (for two-person, interpersonal, synchronous choice): Let G be any two-person game, and let D be a problem that is isomorphic to G with respect to the strategy and payoff structure of the game for both you and the other player, except that the choice behaviour of the other player has been fixed at a certain value that you either know or can reliably estimate — so that what you face, in effect, is a situation in which all that remains to be resolved is your own choice of a strategy. In this case, what you choose in G must coincide with what you choose in D27
However intuitively acceptable this principle is, within the context of ideal games it is subject to precisely the objection raised against the interpersonal separability principle. As the classic Prisoner’s Dilen game illustrates, persons who are disposed to choose in this fashion simply do less well, in a significarntly large class of ideal games, than those who are disposed to reason from a non-separable perspective, and with a view to realizing the gains that can be secured from effective cO-operation. Here, then, is another context within which there is a pragmatic argument against separable and in favor of resolute choice.28
There are a number of plausible extcnsions of resolute reasoning, to ideal games involving more than two players, and to iterated games played under ideal conditions. That is, the logic of the argument is not confined to the two person, “one-shot” case. All of the interactive situations just considered, however, are overly simple in one important respect: there is only one outcome that is Pareto-efficient relative to the standard, non-co-operative, equilibrium solution. What has been offered, then, is at best only a necessary condition of an adequate solution concept for ideal games. What is needed is a theory of co-operative (as distinct from non-co-operative) games, that is, a well developed theory of (explicit and/or tacit) bargaining, for selecting among outcomes that are Pareto-efficient relative to the equilibrium outcomes of the game (or some other appropriate baseline), and which are themselves Pareto-optimal.2 Moreover, for any theory of bargaining that can serve as a normative guide to help people avoid suboptimal outcomes, perhaps the key issue is what constitutes a fair bargain.30
There is also the very important question concerning to what extent the results for ideal games can be extended to games played under more realistic conditions, where players may be uncertain as to the rationality of the other players and where various informational asymmetries obtain. Here, questions of assurance are bound to loom large, even for agents who are otherwise predisposed, as a result of rational deliberation, to co-operate.
However, all of this pertains more to the question of the scope of the argument just rehearsed, and the more pressing concern now is to see what implication this sort of argument has for the problem originally posed — the problem of whether an agent can make a rational commitment to act subject to the constraint of practice rules.
Rules, Resoluteness, and Rationality
I argued in Section 1 that the standard model of rational choice does not provide a secure footing for the rationality of choosing subject to the constraints of practice rules. What I now want to argue is that the alternative model presented in the intervening sections opens the door to understanding and defending a rational commitment to practice rules. Consider first the concept of a practice. One can mark in the abstract concept of being resolute a model for just the sort of hierarchical structure that characterizes practice-constrained choice, both for the kinds of practices that the isolated self may adopt, but also for many of those practices that structure our interactions with others. One has only to observe that a practice can be understood to be a type of plan, and to recall that it is precisely the resolute self that is capable of taking a plan that has been adopted as regulative of future choice, even in the face of what would otherwise count as good reasons to choose differently. But why is a practice to be taken as regulative? Because this is what is needed if individuals are to co-ordinate their actions, or if the isolated individual is to co-ordinate actions over time. For co-ordination to take place, it is not enough that each does what each judges to be “best”; nor is it even enough that each conforms to some rule that each judges would best serve the ends in question, if all were to conform to it. To the contrary, co ordination requires a mutual structuring of activity in terms of a prior, established rule having both normative and positive import: that is, a rule to which all are expected to adhere, and to which it is expected that all (or most) will in fact adhere.31
The rules defining a practice, then, are to be understood as prior to the choices of action that arise under it in both a temporal anda normative sense. There is temporal priority, because what is regulative is a rule that is already established. There is normative priority because the rule takes precedence, at least in principle, over any countervailing choice supporting consideration that can arise at the level of choice of action within the context of a situation to which the rule is applicable.
The logic of practice rules, so conceived, then, involves the notion that one cannot decide to overrule sucha constraint in a given situation to which the practice rule applies by directly appealing to whatever considerations could be adduced in support of the practice itself. Those who participate in such a practice abdicate, in effect, their “right” to make decisions case by case by direct appeal to such underlying considerations.
The sense in which a practice rule is prior to, and establishes non discretionary constraints on choice, is already provided for in the model of resolution choice — in the notion that choice in certain sequential decision problems is constrained by a prior decision to pursue a plan, or a prior (tacit or explicit) understanding as to how choices by quite different individuals are to be co-ordinated. That is, the account of non separable deliberation and choice explored in previous sections provides a model of the kind of intentional co-ordination that is essential to adopting and choosing subject to the constraints of a practice. As I argued at the close of Section 4, the intrapersonal co-ordination problem is resolved by adopting a two-level approach to deliberation and choice. At the first level, consequentially oriented considerations will lead one to adopt a specific plan; and at the second level, the plan that is in fact adopted will set constraints on subsequent choice. In this setting, what is relevant to subsequent intrapersonal choice is not what plan one might have adopted, or what plan it would have been best for one to adopt (by reference to some underlying consideration), but what plan one did in fact adopt. Correspondingly, what is relevant in certain interpersonal decision problems is not what plan the participating individuals might have adopted, or what plan it might have been best to adopt (once again, by reference to some underlying consideration), but what plans are already in place. In each case, then, there is both a positive or “fact of the matter” and a normative dimension to the reference point that emerges for deliberation and decision: what functions as directive for choice is the plan that as a matter of fact has been chosen. This conceptual account holds, it should be noted, even if resoluteness is conceived as merely the imposition, by the earlier self, of a regimen that the later self accepts, or, in the case of interpersonal choice, of a pure convention, among a group of people, regarding how each is to constrain choice in certain situations.
Whether a given practice is fully justified turns, of course, on what arguments can be constructed for the rules themselves. What I have argued is that the logic of interactive (interpersonal or interpersonal) situations is typically such that practice rules are required for the effective promotion of the objectives of the participants. The notion is that there are cases in which the concerns of each cannot be served unless the future is tied down and plans co-ordinated in advance. In such cases each person’s deciding what to do by reference to her own concerns, case by case, will lead to confusion, and the attempt to co-ordinate behavior simply by each trying to predict the behavior of the others will fail.32 When this is the case, one can appeal to the model of non separable deliberation and choice, to show that a commitment to practice rules can be defended pragmatically, by reference to consequences that are assessed from a non-separable, global perspective.
Nothing need be presupposed here regarding what each takes to be the objectives that such a co-ordination scheme is to serve. In particular, there is no reason to suppose that there is some one or more objectives that all participants share. Divergence with respect to ends can be offset by convergence with respect to means by a shared sense that the objectives of each can be more effectively promoted by the adoption of a co-ordination scheme. Correspondingly, there is no need to introduce some ad hoc assumption about persons just happening to attach value to choosing in accordance with such rules. Nor are such persons “rule-bound” in a way that can be criticized from the perspective of a theory of consequential choice.$ The story to be told here can pivot fully and uncompromisingly on consequential concerns. It can be a story of individuals who come to regulate their interactions with themselves over time, and with one another, in accordance with constraints to which time-indexed selves, or distinct individuals, can mutually assent, and who do this from nothing more thana sense of the enhanced power that such a new form of activity gives them with respect to furthering their Own projects and interests,34
I have sought to argue here anumber of things. First, the standard way of thinking about rationality in both intrapersonal and interpersonal contexts unacceptably fails to yield a theory that can render intelligible the notion of having a commitment to practice rules, much less provide for the rationality of being So committed. Second, this feature of the standard theory can be traced back to a basic presupposition of virtually all contemporary accounts of rationality, namely, that consequential reasoning inevitably takes place within the framework of a separability principle. Third, there is a distinct account that renders the notion of a commitment to rules both intelligible and rational: the resolute model. Finaly and more ambitiously, I have sought to show that the resolute model can be defended by appeal to consequentialism itself. The notion is that a consequential argument can be constructed for adopting a more holistic or global approach to deliberation and choice, and this entails, in turn, that in certain cases one should deliberatively suspend the separability principle. In terms of the more familiar notion of practices, the conclusion is that a commitment to practice rules can be defended endogenously, from within a consequentially oriented framework. Alternatively put, the logical structure of intrapersonal and interpersonal co-ordination problems is such that a viable version of consequentialism will be a version of rule consequentialism, in which the notion of a rational commitment to extant rules hasa central place.