Towards a rational application of rational choice theory.

Door: Peter Pappenheim

This post is a first draft. It is published because the author, given his age, is afraid that it might never be published.

Towards a rational application of Rational Choice Theory (hereafter RCT) to social decision-making.







VERBEEK: “The Virtues of Cooperation”.




Half a century of experience and study impressed on me the vast discrepancy between the sophistication of social science and philosophy and the ineffectiveness of their application to social decision-making. Its primary cause is that we do not apply the main faculty to which we owe our technological success: reason. It lead to a book, “The Conceptual Foundations of Decision-making in a Democracy.” (see its web-page on this site), which is written from the point of view of the user of the products of science and philosophy. (Rationality appears briefly in the chapter 3b.2: “Induction: the rationality of learning from experience” which deals with Hume’s paradox.) It must be clear that my concern is not with the current RCT’s. It is their application in the theory and practice of social decision-making that is deficient. This qualification presumes a criterion. For human constructs, a general criterion is provided by the function they are presumed to fulfill. The definition of RCT states that its function is to improve our decision-making. The conclusion of my book as well as of this article is that a condition for the effectiveness of social science, including RCT, is not fulfilled, namely that the theories and statements be presented in a form that does justice to that this function, and that the correctness of their application by the users be monitored. As shown in Chapter 5, definitions do matter.

To be effective, RCT must satisfy certain conditions, amongst which:
- unequivocal and ‘operational’ concepts and definitions (Chapters 2 and 3)
- a complete, realistic and explicit set of axioms (Chapters 4 and 5)


A choice is – or ends – in a decision. In all works I have read, RCT either considers a choice to be rational (a RC) if, when confronted with various action alternatives, the subject chooses that alternative which has the highest expected utility, or calls a decision rational if the action chosen is appropriate to the goal pursued by the subject. Its offspring, the theory of games, deals with a complicating factor: the reaction of other individuals to the decision. In all cases the basic objective of RCT is to determine strategic rules for specific and hypothetical types of choice-situations whose models contain a limited set of manageable variables. For such models RCT develops interesting conclusions. Decision-making pertains to the realm of information processing and is common to all living beings. A criterion for the evaluation of a decision applicable to any decision of any living beings is the extent to which it contributes to the objective of the being, the most general objective being its survival and propagation. The above definitions do not define the rationality of a choice, but its appropriateness. To arrive at an adequate definition, let us look at the two elements of RC.

2a) CHOICE. Choice can be used in two senses. To ‘have’ a choice indicates that there are alternatives, for instance the assortment of flowers in a florist’s shop. To ‘make’ a choice refers to the completed act of choosing, for which we have a unequivocal term: a decision. RCT uses it in this sense; choice here is synonym to decision.

2b) RATIONAL. As used in the definition of RCT, rational can be interpreted:
* As a criterion for evaluating a decision. As there is already a primary criterion for the decision, namely that it is conductive to the objective of the decision-maker, ‘rational’ then must be a secondary criterion, namely a factor for success, for ‘good’. That a rational decision ipso facto must be good, should be justified, but is not.
** As the term which is to be defined. In classical RCT, rational is defined by the property that “the result of the decision satisfies its goal”. As said, that definition then does not define the rationality of a choice, but its adequacy. The implication not only is that a rational decision guarantees success (see *), but that any successful decision must have been rational, an implication which clearly is contradicted by reality: a good choice can be reached by relying on reflexes, intuition or plain luck, while a choice based on perfectly correct reasoning may turn out to be a bad one.
*** To define a specific kind of decision, one that is the result of a decision-making process that engages our faculty of reason. That is the only definition of a rational choice that is unequivocal and operational. It defines a specific class of decisions for which there is no other term and it has discriminatory power because there are many decisions that are taken in other processes such as reflexes or intuition, which may turn out to be better, either because of the time frame (ducking a blow), or because its complexity exceeds the decision-maker’s ability to consider all relevant facts in their proper relationships, in which case he may be better off relying on his intuition.

The correct definition of RC is: A choice is rational if it has been taken by engaging our faculty of reason. The function of RCT then is to increase the probability that such decision-making will reach our objective. The end-quality of the decision appears only as a yardstick of the extent to which that function has been fulfilled.

The functional view of knowledge presented in (*) entails that the evaluation of the rationality of a decision on basis of the decision itself can never be totally objective. Russel had already arrived at the same conclusion: “How do we prove that the man, who believes to be a poached egg, is irrational on other grounds than that he is a minority?” Indeed, believing he is a poached egg would be perfectly rational for a man who wants – by escaping into madness – to cope with the fact that he has just murdered his wife and children. To judge the rationality of a decision requires knowledge of the objective, the motive, of the decision maker. In case of individual decisions, it is usually is not available to the judge of this rationality. The same applies to an evaluation of the utilities of the results of the action alternatives. As long as the decisions only involve the decision-maker, that is his own problem. In social decision-making it engages other members of society. In societies where the decisions are taken on basis of the power of an individual or group, nobody else can do something about it, so judging its rationality is mainly academic. In a democracy, motive is the central problem: to function, social decision-making in a democracy requires that the objective be explicit and mandates that the choice for a specific alternative be justifiable and therefore rational.

Conclusion: ‘rational’ cannot at the same time mean ‘good’ and ‘reasoned’. Any such application is a petitio principii. But that is precisely how RCT is often applied, for instance in economics. This error of argumentation is at the root of most objections to RCT. A theory based on a petitio principii cannot tell us much about our empirical world. Its application will lead to many errors and to conclusions that intuitively will be considered unacceptable and, more important, opens the way to abuses.

2c) THE SUBJECT. Whose reason? A correct application of RCT must specify the entity, the individual, whose rationality is in cause in the evaluation: is it the person taking the decision or is it her observer. Without that specification, an author can – and often does – switch ‘ad hoc’ from one assumption to the other. Intuitively we feel that ‘subject’ should refer to the person taking the decision. If so, as Russel already has pointed out, there is no objective criterion, no means for assessing the rationality of a decision by just looking at it, because we cannot look into the brain of the individual concerned and never will know all the considerations which he took into account. Any judgment about the rationality of somebody else’s decision engages only the rationality of the observer and implies that the objective of the decision (as surmised by the observer) is indeed that of the decision-maker. The observer may have a justifiable claim that the decision is not good, will not be successful, for instance if he has information which the decision-maker did not have. That would not however suffice to brand the decision ‘irrational’, a sin that is fairly common. Another sin is to brand the objective as ‘irrational’, which is incompatible with – as, RCT does – taking the objective as an axiom, as given. Challenging the objective simply generates a new round of decision-making. The subject who takes a decision need not be an individual. For instance in social decision-making it may be a group, but that engages other kinds and aspects of rationality (see for instance Max Weber) and must always be reducible to (some aggregation of) individual decisions.

2d) THE GOAL, THE HIGHEST EXPECTED UTILITY. These terms seem to be unequivocal, but their use has implications. It presumes the availability of the information required for such a judgment, the ability to rank the expected outcome of various action alternatives at least in an ordinal, non-contradictory way. RCT also implies that ‘utility’ and ‘goal’ can be defined with sufficient precision to allow extension of the findings of RCT to the actual situation in which it is to be applied. ‘Sufficient precision’ involves for instance that all relevant and possible results of the decision are taken into consideration. The utility or goal must be independent of these considerations and not biased by the objectives of other individuals: the facts used in the determination of the expected results must be obtained in a process that is as objective as possible.

2e) UTILITY VERSUS FUNCTION. Conflation of these terms is a major source of errors in argumentation, especially in the discussions of the use of RCT as an axiom in economics. Utility is the value that an individual attaches to the (expected) results of an action, while function refers to the appropriateness of an action alternative to achieve these results. Decision-making involves two activities:
- How to rank the expected results in terms of utility. The evaluation of the utility of a result always is subjective.
- How to achieve these expected results. That is an activity that can engage reason, which is a function (often in the form of science) which is amenable to a certain degree of objectivity.

In the fictional examples used by many RCT’s, there is only one (often simple) known ranking of the expected results of all alternatives in terms of satisfying the objective, of utility, which implies that the appropriateness of the action chosen is known. The distinction between utility and function then is immaterial because both function and utility have been discounted in that ranking. In the practice of social decision-making, the condition of one simple and known ranking is seldom satisfied, and in a democracy the two aspects should be separated.

Conclusion: In decision-making, ‘rational’ in RCT should be applied to the process of determining the expected result of the action-alternatives considered, it evaluates their functionality in reaching the result with the highest (expected) utility.


3a) WHAT IS REASON? The above definition of a rational choice is incomplete unless we specify what is meant by ‘reason, rational’. Classic RCT does not provide an ‘operational’ explication of what reason consists of. I must then fall back on Webster, whose definition of ‘rational’ is derived from “the ability to reason”. Reason as a verb is defined as: “to think coherently and logically; to draw inferences and conclusions from facts known or assumed; to argue or talk in a logical way; to analyze; to think logically about; to think out systematically”. Using the key words particular to reason and rearranging them to form a valid sentence, we get: “rational (= to reason) is to draw inferences/conclusions (= to deduce) and analyze in a coherent, logical and systematical way.” Reason is the ability to deduce and analyze in a coherent, logical and systematical way.

The concept of reason and its definition must be congruent with the current stage of our knowledge about the human being in general and the functioning of his mind in particular. The terminology used must enable us to integrate that concept into that knowledge. That job presumes a shared, up to date, model of a human being. Social science and philosophy have not yet even acknowledged the necessity for such a venture. So we must do the best we can with the knowledge we have and which is the realm of psychologists, biologists, system analysts, neuro-physiologists etc. plus – mainly as coordinators – philosophers who are interested in the subject, especially methodology. Clearly, nobody can do the job on his own. But somebody must put the bell on the cat. The required current knowledge about life is available to anyone with a higher education and it can provide at least some basic concepts and conclusions that are sufficient to put the application of RCT on the track. Below a brief summary; a more detailed exposition can be found in part two of the book.

The main feature that distinguishes a living being from an inert item is the self-directed use of information. The major steps in the development of the living world are embodied in the means by which the being concerned ‘learns’ from experience to adapt to different circumstances. To that end, they must in any case have a memory to record the experience (both the action and its results) and – if the sort is to learn from it – they must transmit it to others. The lowest level (bacteria) can learn only as a sort by recording the successes in the genes of those who survive and reproduce. The next level (animals) can in addition learn at the level of the individual through a central nervous system, a brain, which contains specific information-processing features and a neural (and thus internal) memory. Because the record in the neural memory disappears with the individual, they still can learn as a sort mainly through the genes (though some forms of behavior can be transmitted through teaching and imitation from parent to offspring). The highest level consists of human beings. In addition to the previous forms, they can also learn as a sort from the experience of individuals, they can pool the experience of both past and current human beings by creating and using a common and powerful means of processing and communicating it, primarily a symbolic language, and by external memories: that of fellow humans, books etc. They have introduced a new phenomenon on earth, our culture, and thus created a totally new world: art, entertainment and the most prominent and powerful means for our survival and propagation: science. (Whatever the shortcomings of his theory and of his personailty, the credit for uncovering this new world (he called is noosphere) as the essence of humanity goes to Theilhard du Chardin).

‘Pooling’ implies cooperation, that man is a social being. There are few living creatures that are more dependent on their society than modern man. Deprived of his culture, he is not even a naked ape: besides the protection of fur, he lacks for instance muscular strength, an adequate sense of smell and probably some instincts telling him what is digestible and what is poison. Human culture requires a correspondingly powerful information-processing organ: a human brain and the capacity to supplemented his memory is by external memories (books, photos and now computers etc.) which provide a practically infinite storage room. But just storing a lot of information does not explain is current dominance; it also must be adequately processed. The basic feature that distinguishes human information processing from any other being is his ability to manipulate his knowledge, to generalize, to ‘look’ at his own information processing (reflective thinking) and correct it (the main function of reasom), and through imagination to create a whole world of virtual information.  Reason – as defined above – is a specific application of the ability of reflective thinking.

3b) WHY IS REASON ‘GOOD’? Animals learn from experience, from the results of their actual choices, but pay a high price: the cost of failures, often death. Imagination provides a ‘virtually’ unlimited supply of alternatives, and reason supplies a means for eliminating those which cannot further our objective and to make a provisional selection of the one most likely to achieve it, at no cost other than this information process. RCT assumes that the application of reason to decision-making will promote the probability that the alternative chosen will satisfy the objective of the subject, that the action we decided upon will maximize the utility. The ability to deduce is directed by logic, and the basic assumption underlying logic is causality, is the rejection of miracles as an explanation of events or as a basis for predicting them. In the realm of reason, ‘miracle’ is replaced by ‘we have not yet found an explanation’. In the end, as with all features of living beings, the justification of logic is the experience of its application: man has survived and prospered by relying on it. The almost universal acceptance of logic and its off-spring, mathematics, in all developed countries is a tribute to the success of their application: their rules have never been shown to be wrong by any event which can be replicated, while everyone can at will create experiments which confirm their correctness. We can put the issue to rest by excluding anybody who contests the validity of logic from the discussion, and we can do so with a clear conscience, for such a discussion always is an exercise in futility.

The categories of reason, causality and logic serve as control instruments for eliminating incorrect decisions before they have been implemented. Reason is good because – as the ability to deduce – it enhances our ability to exploit our past experience as a means for increasing the probability that the result of current decisions will meet their objective. But it is only one of the factors of a good decision.

A decision based on correct deductions then cannot ipso facto be qualified as good. The objective may be bad (engaging the category of justice, morals) or the assumptions about the facts used in the deduction may be wrong. Also the previous decision to engage reason may be wrong if – due to time pressure or insufficient information – we should have relied on reflexes or intuition. But if applicable, reason can improve our decision-making yet never hurt it; it cannot however ensure that the decision is good. From the conclusion that a decision is rational we can only conclude that the process of deduction has been correctly applied.

3c) IRRATIONAL: NON-RATIONAL OR ANTI-RATIONAL? To deduce any normative statement about the application of reason and to maximize its empirical content, we must compare it to its alternative, irrational, which – in the context of correctly defined RCT – means: “not according to the prescription of reason”. That definition is bivalent.

It can mean ‘non-rational’ (without engaging the faculty of reason). If so, the rules of logic state that one cannot deduce any further statement from it (ex falsibus omnii). Specifically, we cannot a priori attach to ‘non-rational’ any pejorative connotation.  By itself, non-rational is a value-neutral concept. We may have had good reason to rely for instance on intuition instead of reason, usually because the conditions for applying reason are not fulfilled. In fact, the large majority of actual decisions fall into the category of ‘non-rational’, see (*), Part 2b, “Parallel versus sequential information processing” . Usually, non-rational decision will turn out to be good, at least if the individual is to survive. A first shortcoming of the application of most RCT’s, especially in economics, is that they ignore non-rational decisions.

The pejorative connotation is applicable only to ‘anti-rational’: reason has been applied but the resulting choice is not followed. We know that smoking is bad but we still light a cigarette. An even more interesting example is overweight. For although its nefarious consequences are by now general knowledge, research has led to the conclusion that it is practically impossible to motivate more than a small part of the population to change their eating habits. (Kant has a word for it: heteronomy. I will not use it for – to do justice to the current knowledge about man – we must assign a different content to its opposite, autonomy, than he did). Only anti-rational decisions can be branded as bad. Their ‘badness’ however is not due to any deficiency of reason, of the informational part of the decision-making process. The badness of anti-rational decisions is caused by the inability of the subject to execute what reason has shown to be the best choice. The RCT-alternative then was not really a feasible alternative. That does not however engage the rationality of the ‘best’ choice, only its translation into a decision. (Very often a motive such a status is branded branded irrational, but in fact that is a sin against reason by the observer).

Summing up: we have three types of decisions, classified according to their rationality and qualified according to their expected contribution to the objective:
- Rational decisions (good)
- Non-rational decisions (depends on circumstances)
- Anti-rational decisions (bad)


The basic RCT consists of deductions (from the definition of RC) that are intended to be valid for all RC’s, given a few axioms about the structure of the decision-making process, for example the prisoners’ dilemma in the theory of games. Given the objective, the conditions that an RC must meet to achieve it are:
1) The application of reason must be appropriate, and the subject must be able to do so
2) The set of alternatives must contain all alternatives that could lead to the objective
3) The statements of facts engaged in the reasoning process must be correct; that includes the expectations of the events resulting from the actions chosen.
4) The values, in terms of the objective, attributed to these events must be correct (the job of RCT)
5) If it is to result in a rational decision, the subject must be able to effectuate the action chosen.

To apply RCT to ‘real’ situations (to derive synthetic statements from it), we must define the field of reality to which the variables in the specific RCT correspond (we must give them empirical content). What is the mechanism of reasoning? How do people obtain and rank their objectives? How do they determine the expected utility of their actions? What information do they need or use and how can they get it? What is the structure of a model based on RCT, especially the interrelationships between the variables: the objectives, the means and the information about facts? These elements differ from one situation to another. To be at all relevant, a theory about decision-making must take account of at least all those factors which are of such importance that neglect of them would lead to a different decision. Condition 4 is the job of RCT, which is applicable only if the other four conditions are met. Even if the system is very complex but all independent variables consist of inert items, as in Operation Research, the application is straightforward and its problems allow unequivocal solutions.

In other fields of decision-making such as economics, RCT can provide a structure and axioms. To that end, the ‘reality’ of that specific field is reduced – through an appropriate choice of axioms  – to a model that allows application of RCT to produce a theory (for instance of economics) which at this stage still is analytical. If such a construction satisfies the requirements of argumentation, then it is ‘true’ by convention, for instance in case of economics the equalization of marginal utilities. RCT has been very productive in this respect. The problem mainly is with their application, namely ensuring correspondence of the axioms to the reality in which the decision is taken.


Because a basic rule of information processing is: “garbage in, garbage out”. The application to actual decision-making of RCT and of theories using it as an axiom is subject to all five conditions mentioned above. It is precisely failure to adequately define and consider all these conditions that is the Achilles heel of many applications of RCT and thus of the ineffectiveness of social science and philosophy using it. The petitio principii implied by the classic definition of RCT which equates ‘rational’ with ‘having the highest expected utility’ or ‘being appropriate to the goal pursued by the subject’ allows unfulfilled conditions to be swept under the carpet.

Both logic and common sense tell us that condition no.3, obtaining relevant and reliable knowledge about the expected results of the available alternatives, has priority over evaluating them in terms of the objective; it is a precondition for applying any RCT. In the absence of such knowledge we are better off by relying on intuition or tossing coins. Condition no.3 – ensuring the correctness of the statements about facts which enter the decision – is also the condition which causes most of the trouble and is the most difficult to meet; it can only be met if the definition of the terms of the statements which refer to these facts are defined with sufficient precision.

The applicability to actual decision-making of any theory thus depends on these definitions, whose suitability will emerge during the ‘axiomatization’ of the theory considered. Axiomatization thus is the most cost-efficient means of control and one that is always possible. An – often willfully – ignored property is that axiomatization does not require much knowledge about the specific discipline of the theory investigated. Given the proficiencies of their academic discipline and their position at the apex of decision-making, game theorists, “besliskundigen” in my language, are uniquely positioned to monitor the application of RCT by other disciplines. They are well aware for instance of the property (holistic) of social systems mentioned in the introduction but studiously ignored by much of the academic organization.

Assessing the correctness of statements about all factors entering the decision is itself a decision, which again may require a new round of decision-making, ending in some fundamental statement which we will accept as correct either because consent on it is unanimous, or because it is substantiated by an acknowledged authority. In social decision-making, the set of statements about facts must take account of the interrelation between the organization of society and the objectives and motivations of the individual. Verbeek’s work (see below) for instance argues that – in the absence of reliable information about the behavior of others – trust is a reasonable ‘default value’. Trustworthiness being of benefit of all, the ability of society to generate trustworthiness must be taken into account when judging the performance of a specific social order, for instance a capitalistic market economy; economics we must consider its effect on trustworthiness as one of its ‘external effects’.

An immediate result could be achieved by a correct application of RCT in the field of politics. Almost daily my newspaper publishes an article by some ‘authority’, usually a professor, who convincingly argues that an actual or proposed decision by a politician is wrong or that he is lying. Worse, ever so often that turns out be correct. No wonder that the general appreciation of politicians ranks amongst the lowest of all legitimate professions, that they are considered incompetent, unreliable and power-obsessed if not corrupt. But the very fact that their positions are discussed and must be justified indicates that they are intended to be rational. And, as explained, we must assume that the decisions that politicians take are rational, are not taken without engaging their faculty of reason. Unless we assume that politicians are born as power hungry pathological liars, other factors than rationality must be the cause of their low status. In a democracy, we must assume that their behavior mirrors the morals and intellectual capabilities of its citizens. Discussions about the functioning of our democracies mainly involve the organizational system (direct democracy, proportional representation, district system, etc.) notwithstanding the fact that about all conceivable systems which are feasible in a modern society are or have been tried without noticeable effect on the disaffection with politics, which remains general; only the accents vary with the system. A rational user of RCT therefore must conclude that other conditions must be in cause, and that the correct approach is to apply RCT to politics in the same we have adopted with economics, replacing dollars by votes. (I know only one such attempt: “Demokratie en welvaartstheorie”, by H, v.d. Doel, 1980. Unfortunately the then still young author suffered a disabling stroke, thus robbing us of a brilliant and honest mind.) Anyway, his book mainly deals with just one of the many aspect of the problem: a party’s choice of its position within the political spectrum.

Today’s democracies are party systems. On any given decision, we cannot assume just one given objective, we must take account of two rationalities, a fact already noted a century ago by Max Weber. The individual has just his own, but the politician has two: functional (the party) and substantial (society) rationality. In a democracy, substantial rationality ideally should be a correct aggregation of the objectives (rationality) of the voters. And it must ensure that the politician’s functional rationality is compatible with is above-defined substantial rationality. These conditions pertain to the field of the organization of social decision-making; the first one to political philosophy/science and the second to political science/management/administration. They must be considered when writing a realistic job-description of politicians, and only with such a description can we adequately judge – and hopefully improve – their performance. I am confident that this will show that we expect them to perform functions which are not their primary task and for which they are not qualified and cannot expected to be, such as developing a vision or establishing facts. Or leadership: in a democracy they can only exercise it within  the organization for which they are responsible, a government or a party; but not for the whole society. That rises the obvious question: if not the politicians, then who must perform these functions? Part of that answer forms the conclusion of my book and will be summarized in a paper yet to be written, and some of it is found in chapter 7).

Verbeek’s doctoral thesis “The Virtues of Cooperation” (1998. Commercial edition in 2002 by Kluwer, titled: “Moral Philosophy and Instrumental Rationality”.)

It is selected here because it deals with the problem of ‘conformity to norms’ which was not solved by Ralws and because it illustrates that the purely instrumental rationality of RCT must be supplemented by other considerations if it is to serve as a realistic apprehension of norms in a well-functioning, democratic society. Also, the importance of cooperation is often ignored in social philosophy. The next,  introductory paragraph is mine.

The coexistence and cooperation required for any viable society implies a certain level of coordination of decision making which inevitably limits the range of behavior of its members that is tolerated by that society. Some actions are prohibited and others are mandatory. With animals, the agent ensuring that behavior is instinct or the power of a leader. An animal’s instinct is self-enforcing: the animal cannot act otherwise. In a fight for leadership, the winning A-wolf cannot bite the throat of the younger loser who offered to to him; he probably cannot even conceive such an action and certainly does not know that only his instinct prevented him from doing what might have been in his best interest. Herein lies the ineffectiveness of instincts and problem of norms if the being is endowed with the faculty of reason. If the decision required by norm or instinct is at odds with the one that satisfies his own objectives, why should a creature endowed with the capability of reason decide to comply with that instinct or norm for any other reason than the power of an enforcer who renders deviating from the norm less advantageous that conforming to it? That question is especially important if – as I do – you value democracy, for a democratic society (which should be a voluntary cooperative venture) cannot run primarily on coercion. Bruno Verbeek has tackled precisely that question in his book “The Virtues of Cooperation” and done so in a way that meets the basic rules for argumentation in a democracy presented in my book. I will try to present his conclusions in a nutshell; any errors are to be put on my account.

Back to Verbeek. He uses ‘rational’ in the first interpretation (2a) which I think is not the right one, but his findings are just as relevant if we use my interpretation, provided we limit the scope of the conclusions to decisions meeting the conditions for the use of reason. A norm only makes sense if people can deviate from it. Some norms (such as the standardization of nuts and bolts) are self-enforcing. I agree with Verbeek that we can ignore them because conformity to such norms is unproblematic. The norms relevant to his and my problem are those whose compliance entails a sacrifice by the complying individual, as they might and usually do prohibit actions which could bring him a higher advantage than those allowed.

Any reasonable person will understand the necessity of norms, certainly if compliance to norms by others is to her own benefit. She will thus support the establishment and enforcement of such norms by society. From the point of view of RCT, the two main problems of such norms are:
1) How do we ensure that rational people support the establishment of all the norms necessary for a viable society or otherwise are to the advantage of all?
2) How do we ensure that they act in accordance with the norm in cases where a personal, specific, objective/interest would dictate another action?

ad 1) Classical RCT holds that the norms must – if all are to comply – be to the benefit of  all. It considers the advantages of the norm as given for the individual concerned. Because the benefits of a norm diminish – possibly in geometric proportion – with the incidence of successful infringements, classical RCT puts its emphasis on the second problem.

ad 2) Classical RCT sees sanctions – which make deviance more costly than the benefits which the individual can expect from it – as the ‘natural’ means for ensuring compliance, and therefore concentrates on increasing the costs of deviance. Verbeek notes that if we ensure compliance only by increasing the cost of deviance through sanctions, that endeavor will be successful only if we can thus prevent practically all deviance, for – as said – successful deviance decreases the advantage of compliance. If deviance is more than the result of the inevitable but incidental irrational behavior, the positive net advantage of the norm must be guaranteed by increasing of the costs of infringements. The above ‘if’ translates into severe sanctions and strict control, both of which are social costs that in turn decrease the net benefit of the norm to society. A society where compliance to norms relies exclusively on the purely instrumental rationality of classical RCT would be some kind of repressive police state; it will not be a viable society and certainly be an undemocratic one. Verbeek calls it “perfect tyranny”. (author: In addition – while it may be successful in preventing most negative behavior and thus can ensure coexistence – it is inadequate for generating the enduring and extensive cooperation on which our western societies are based.) Verbeek also mentions another (I think Hobbes’) problem of such a purely instrumental and individualistic application of reason: the punishment of deviant behavior itself must be administered by a higher-level institution, whose members may make the same calculations and must therefore be controlled by a still higher institution etc. Each higher level will contain fewer individuals, but the negative effect of a single infringement at that higher level on compliance at a lower level affects many, cumulating the costs.

The major contribution of Verbeek is that he notes that applying such a view of RCT to norms is justifiable only if the advantages which the norm is supposed to generate do not require cooperation of other individuals, meaning that they do not just forbid certain actions, but mandates the benefical ones. He specifically mentions collective goods that are also a major subject in my book. Classical RCT already has shown that if two or more people are involved in a decision, the best choice for everybody can be different from what everyone would have chosen on his own. Take the standard prisoner’s dilemma. Two suspects of a crime are put into separate cells and each is given the option to confess and incriminate the other, resulting in a six-month prison sentence if the other also confesses. If the other does not confess, he will get scot-free. If he does not confess, but the other does, he will get six years. Obviously the best choice is that neither would confess, for both would go unpunished. But that alternative is out of reach if each of them thinks that there is a good chance that the other will confess. The best choice therefore depends on the expectation that the other will not rattle. As reason for making the best choice, classical RCT relies on the fear of revenge, a solution that implies that both are confident that the other will fear retribution more than he fears the extra prison sentence. That is the system of the mafia, which illustrates its shortcomings. Even the death penalty enacted by the mafia did not prevent a few treasons, and most of us would anyway not want to live in such a society.

A viable and democratic society presumes that its members have a predisposition to act cooperatively, that they see the advantage of a general expectation that others will follow the norm and that those who do not, will be punished. Verbeek investigates what these predispositions are. A first and foremost requirement is reciprocity:  prisoners will only keep mum if they expect that the other will do the same, the same applies to the other. Without at least the expectation of reciprocity, collective goods will not materialize. So cooperation requires another element: they must trust the others to act reciprocally, to be reliable, to be trustworthy. Reliability is in a sense an objective property of a person, often tied to a specific situation: she will act in a consistent way in that situation because in the end it is to her  advantage; it is objective if it is based on a record of such consistency. That does not apply to trust: a trustworthy person acts that way because she is trusted to do so, period. She understands the advantage that comes with being trusted by others. However, trust is not conditional on knowledge of the other. We may trust a person to keep her promise if that is what we generally expect people to do. Summing up: A reciprocal actor acts cooperatively when he has good reasons to expect others to cooperate; a trustworthy actor acts cooperatively when he is trusted to do so, trustworthiness is conditional on trust. Trust is what makes cooperation possible in situations where ‘pure’ RCT would fail. Trust is the ‘default value’ in the absence of knowledge about the other, it is what we assume whenever we cannot ascertain the reliability of the other parties. (Trust therefore is not a purely individual thing, it presupposes a society which promotes trustworthiness, author).

Suppose the prisoners have promised each other never to rattle, there still is no way to be sure beforehand that none will do so, that the other is reliable. They must trust the other to keep his promise. Without that trust it is impossible to reach the best solution. Trustful agents have the capacity to “make the first move”, trust makes it possible to cooperate successfully with trustworthy agents; as the saying goes: there is honor even amongst thieves. Trust is a cooperative virtue. But is it rational? Verbeek shows that there are good reasons to answer ‘yes’, one of them being that in a healthy society most people ipso facto are trustworthy. (A worthwhile subject of investigation would be the relationship between the nature of a society – as embodied by its institutions – and the level of trust it generates, author).

Fairness is another virtue promoting the compliance to norms: common sense tells us that people are more prone to comply to a norm which they acknowledge to be just than to one they feel to be unjust.  Fairness also is a condition for another, second-order, condition: resentment. Verbeek deals mainly with a specific case of unfairness that he calls “exploitation”.  He shows that this fairness partly overlaps with the principle of reciprocity and provides a theoretical justification of fairness as a perquisite for compliance. But he begs the question: what is fair? That must be explicitly defined because this virtue only works if – when dealing with a specific norm – a large majority of people share that concept of fairness. Verbeek seems to rely mainly on Rawls, but – by his own admission – Rawls’ theory of justice does not solve Verbeek’s problem of compliance.  (My book proposes a theory of justice which solves that problem.)  Given his objective, Verbeek can neglect that problem: he points to the fact that usually people do have a similar sense of fairness (author: experience shows that even if they do, their conclusions in specific cases often are different). Total unanimity is not necessary. Verbeek, p166, quoting from Sugden, writes: “…a cooperative virtuous person will tolerate a certain a bit of parasitic behaviour from others while cooperating. … What the example brings home is  that there is a threshold of non-cooperation beyond which reciprocally disposed agents are no longer under a duty towards each other  to cooperate.”

Resentment (as part of a more general disposition to express disapproval of lack of  cooperative virtues) is a powerful discouragement of deviating behaviour. As it presupposes these cooperative virtues, it is a cooperative virtue only in a secondary sense.

Are cooperative virtues rational? Is it rational to be trustworthy and fair in all decisions we have to take? That is the subject of the closing chapter of Verbeek, but I will present the subject in a different, more general form. Clearly, when considering individual decisions of a person, it is rational to evaluate them purely on their contribution to the person’s choice; moreover, the above assumption of knowledge of that contribution is usually unrealistic. In actual decision-making the choice always precedes the action taken and therefore implies a prediction of all consequences of that action. To be qualified as rational, all these consequences must have been considered, which usually is difficult and often impossible. If that is impossible, or if the individual often faces decisions presenting the same decision-making problem, then a general disposition, a policy, a strategy covering all these cases, is rational whenever – over the totality of all choices considered – the expectation of the net total of advantage form all such decisions is higher than without that disposition. In the case of compliance to a norm, Verbeek shows that the predisposition to comply implies that we have reason to expect that others share this disposition. This condition is not absolute: it can be rational to act virtuously even in the absence of such certainty, but only exceptionally: if we did so consistently, others would notice this and – following RCT – exploit our disposition to their advantage as long as they can get away with it. Classic RCT does not by itself provide any foundation for the assumption that others will act cooperatively if we do, because it still assumes that compliance of others is assured only by sanctions. As long as we remain within the framework of defining RC by its success in achieving our personal objective (Verbeek calls this pragmatic foundationalism) there is no way out, we cannot explain how a livable society could emerge if its members choose rationally. He concludes:

“From this follows my main objection against pragmatic foundationalism. I take rationality to be a cognitive criterion. That is, a decision ‘D’ is rational not because it happens to render the best results. Rather, ‘D’ is rational if, and only if, it is made for the right reasons. Amongst those reasons may be the expectation of pragmatic success. If an agent is rational, she will do well in terms of the evaluations of her options. But not the other way round. If an agent does well, that is no warranty for her rationality. Pragmatic success is at best an indication of rationality, not a clear and distinct proof of it.” That conclusion is similar to the one deduced in the introduction of this paper from the analysis of the definition of RC. But Verbeek’s criterion “made for the right reasons” still does not meet the requirements of a workable definition: unless further defined, ‘right’ has no discriminatory power: it can refer to the moral aspects and thus belongs to the objective, or to the use of reason, specifically the rules dictated by it, for instance of deduction. As in the classic definitions of RC the objectives are a given, it can only – as concluded in chapter 2/3 – refer to the use of reason. As said, the deductions of Verbeek’s cooperatives virtues are not affected.

Without cooperative virtues, a society of rational individuals is doomed to anarchy or to the tyranny of the power to punish. Cooperative virtues are dependent on reciprocity, on the extent to which other citizens exhibit those virtues by observing the norms of that society. It is reasonable to assume that a police state also is detrimental to cooperative virtues. Deviance must – by sanctions – be limited to the threshold-value of erosion of cooperative virtues, which can only be achieved without extreme sanctions if cooperative virtues have not deteriorated too much through either deviance or repression: the model exhibits an unstable equilibrium similar to that of bicycle-riding. As soon as we want to apply Verbeek’s work in actual social decision-making, to derive from it synthetic statements, the question arises: where do these cooperative virtues come from? As shown, they cannot be based on purely individual RC. Observance of norms by a large majority of other citizens also cannot be a sufficient cause for an individual to comply. The required ‘internal’ dispositions of an individual are the field of investigation of (social) psychology that will show that many of these dispositions are culturally acquired. Being one of the most pervasive influences on our dispositions, the way we have to earn our living must be considered a prime factor. The predisposition thus becomes an external effect of the way our economic life is organized which should not, but always is, ignored in mainstream economics. It could be argued that the main effort of business organizations is directed towards reducing competition and preventing the customer from using his reason, and that this is necessary to keep the system as it is from collapsing.

Part of Verbeek’s book also contains a presentation of Sudgen’s classic RCT’s explanation of the emergence of norms. While that is largely irrelevant to the problem of compliance, it is an example of a misconception that has wrought havoc in various political philosophies and therefore is dealt with in the appendix.


7a) INTRODUCTION. If you consider today’s decision-making in our democracies basically sound, you do not belong to the intended audience of my work. The organization of today’s western democracy is a product of responses to immediate contingencies rather than of the application of reason which, at the time of the Enlightenment, lay at its  foundation. The tools of reason in this field are mainly social philosophy and science. The choices in these disciplines also pertain to the field of social decision-making. They are intended to be rational, but not as response to a dictate by some superhuman authority or law of nature, but as a condition for being effective. To that end they must satisfy the above five conditions for rational application of RCT mentioned in chapter 4.

One in particular, point 3 (relation of means to results) needs attention. It mandates that we consider all the properties of a system such as a human society and that we take account of the latest knowledge about it, especially its structural, mathematical, property (which for lack of a better word I have called holistic). It is:
- complex
- interdependent (recursive, circular etc.)
- dynamic
- stochastic
- its elements can learn from past applications.

The first two alone are sufficient to require that such systems be treated as whole; the next two guarantee disaster if that is not done, and the fifth limits the predictive power, the control, which any theory could give us over such a system. In case of social systems it invalidates the end-state (utopian) approach.

Dealing with the all of the variables and their conditions (see chapter 4)  implies the findings of other disciplines such as biology and psychology. Within social philosophy and science, it requires coordination of their various sub-disciplines. If the findings of a discipline are intended to be used as input for rational decision-making, that imposes conditions on these findings which are not met today and which are the main subject of my book.  Science is a cooperative venture between scientists and those who use their products, which may – as explained – include other scientists who use them for their axioms . While in my book another road is taken, I will below rephrased as far as possible in the terminology of Verbeek’s book.

7b) FUNCTION AND UTILITY OF SCIENCE. What distinguishes a living being from an inert object is not the ability to act on its environment; sunrays also do. It is its ability to direct its actions towards its own objectives, usually survival and propagation, by choosing amongst available action alternative the one which is most likely to contribute to its objectives. It has at its disposal various items (organs, instruments etc.) which can perform such actions. The ability to perform that action is a function of that item. In the paragraph “Utility versus Function” (Chapter 2e) it has been noted that the concept of function is different from the concept of utility, and that the evaluation of the function of an item can achieve a ceratin degree of objectivity. Science does perform a generally acknowledged function of the highest importance in decision-making: to provide the knowledge for condition 3, establishing the estimate of the events resulting from effectuating the choice considered. That function should serve as a major objective in the decisions of scientists if their findings are intended to play a role in the RC’s of the users, which range from individuals, including other scientists, to mankind. As said in the introduction, this paper takes the point of view of the ‘consumers’, of the ‘substantial rationality’ of the venture of science and related philosophy as seen by social-decision makers. (Remember, the utility of scientific knowledge is not in cause, only its function in social decision-making However, there is a link: without its function of prediction, science is useless for most people which then might balk at paying scientists for their efforts.)

Following Verbeek, smooth functioning of science then implies two basic cooperative virtues: trustworthiness and fairness. Trustworthiness implies that the users of science must trust the producers to ensure that their statements are as ‘true’ as they can make them. That entails reliability as a second order consideration which can be tested. Fairness requires that the compensation which scientists get for their products is commensurate to their usefulness (to the whole of present and future society), a subject which belongs to the organisation of society in general and economics in particular and exceeds the scope of this paper. Anyway, fairness is a secondary consideration because if the product does not fulfil any function, there is no cause for exacting any cooperation.

7c) THE COOPERATIVE VIRTUE OF TRUSTWORTHINESS IN SCIENCE:. A substantial part of my book deals with the reliability of scientific statements and how to ensure it to the best of our ability. The current knowledge about life and information induced me to take a different view than Popper on some subjects, especially on the possibility of a totally objective evaluation (in his case falsification) of scientific theories and his notion of an observer-independent existence of world three objects. But it also supports his conclusions that achieving the maximum of objectivity is a precondition of correspondence of statement to fact, and that its ex-post success or failure is a factor in its trustworthiness. The pursuit of objectivity in establishing and evaluating scientific theories is widely acknowledged by scientists. In my book it is shown that maximum objectivity also is mandated by democracy for reasons of fairness, for  the introduction of unnecessary subjective elements by the supplier of facts may surreptitiously substitute the objectives of the supplier for those of the user.  Objectivity in the statements of scientists then engages both cooperative virtues. (Because fairness leads to the same practical conclusions as trustworthiness in the evaluation of the products of science, we will limit the discussion to trustworthiness.)  Trust is a substitute for knowing reliability, namely whenever the prevailing conditions do not allow its assessment. Reason tells us that we should perform the check only if the damage to be expected by applying a false statement is substantially higher than the costs of checking its reliability. Such a check however has another advantage: it provides a measure of the rationality of trusting other statements of that scientist which we cannot check:
- If it is difficult to check these other statements, then evidence of the falseness of the statement considered or its lack of objectivity reduces the trustworthiness of its author.
- If the statement is part of a theoretical system, then its falseness puts the whole system in cause.

Concluding: checking the validity, c.q. objectivity, of scientific statements is mandatory if science is to fulfil its major function.

We have two instruments for checking scientific statements: tests and axiomatisation (examining the inner consistency of the statements). Both contain conventional elements because they require the acceptance of the rules and norms of logic and mathematics and of the truth of the initial conditions, which rarely if ever can be proved even when concerning the inert world.  For most living systems, certainly a human society, we sometimes cannot, even in theory, design such a test. If we can, most of these tests are impossible to execute in practice or at least very costly and time-consuming. The second means, axiomatization, always is possible and can be done by one person, often in a few hours or weeks. Also, a single failure to meet the requirements of axiomatization is sufficient cause for rejecting the proposition as being unable to allow any conclusion (de falsibus omnii): it has no discriminatory power. In the application of scientific theories, RCT therefore mandates axiomatization as the first step in evaluating their ability to fulfill their major function, to help us in decision-making. It is disconcerting to see how often that requirement is not met. Many well-known theories and philosophies (and conclusions derived from them) are based on an error of deduction or contain other breaches of the elementary rules of scientific argumentation, such as lack of explicit and operational definitions and of their consistent use throughout the work..


It is closely related to trustworthiness but less apparently less obvious. The application of a theory always has the form: “if… (a set of axioms)+(rules of deduction) then… (conclusion)”. Many of these axioms consist of findings from other disciplines. Both reason  and trustworthiness mandate that the basic input of facts as axioms in the system, the starting points, should be whatever information about these facts we can gather today, not what predecessors, however famous, have written about them (and even less, what best fits the theory under consideration). That implies not only that this knowledge can be found,, but also that is acknowledged to be the best that is currently available, a condition which often is not fulfilled.

Rawls and Nozich are two examples of the consequences of breaching the rule of realism in favor of starting from existing theories. Rawls (A Theory of Justice) intends to present “…a theory of justice that generalizes and carries to a higher level of abstraction the traditional conception of a social contract”. I used his book as the main basis for my work on justice as it was the most up to date, systematic and generally acknowledged work on contract theory that is the only rational theory for a democratic society, as intuition and reason told me long ago. By his own admission, Rawls failed to solve the problem of compliance, a basic problem that must be solved in a democracy. So I had to figure out why he did not solve that problem. Consistent with the above, I investigated his axioms and confronted his approach with what I knew about the development of human societies. I read the books of other exponents of contract theory, also Rousseau’s “Du Contrat Social”. For various reasons Rousseau’s work cannot serve as the basis for a modern social philosophy. But I retained an element which was rational and still valid today, namely that you cannot simultaneously define both the type of society you want, and principles of justice which are supposed to regulate it. One of the two has to be given priority. Rousseau has chosen for society and sees justice as subordinated to the social contract. Both logic and history justify that assumption.  For the virtue of realism would lead us to start an investigation of contract theory by observing what a contract ‘really’ is, how it is used in daily practice. It is the formalization of a voluntary agreement to comply to certain rules. These are deduced from a shared objective which is the first statement of any contract. Contrary to Rawls, I therefore started with a definition of that objective, with what I consider a well-ordered society, namely democracy. I then derived the principles of justice from that definition. I can then do without questionable concepts such as veils of ignorance or a state of nature. (Let me repeat that without Rawls’ book it would have taken me far longer to write mine. I still cannot understand why any real scientist or philosopher would consider, as some did, that my attempt shows a lack of respect and appreciation of Rawls’ work. I would feel honored if someone would take my work seriously enough to do the same job on it.)

A second example is Nozick (Anarchy, State and Utopia). He felt obliged to start from the writings of famous predecessors, mainly Locke and his state of nature. But he ignored a crucial condition that Locke put on the justice of original acquisition: “that his labour be mingled with it”. He also ignored the fact that ordered and viable societies existed before we started to discuss things as abstract as the principles of justice. Most important, he underplayed the role of cooperation and put his accent on protection. His minimal state is already contained in these assumptions.  He proposed his own line of arguments only as an afterthought in the short, third and last Part (“A Framework for Utopia”), and then only to lend more appeal to the minimal state which he had justified in the first parts. The approach of this Framework for Utopia closely parallels mine because it starts from a similar principle: that everybody should have the right to his own utopia. Unfortunately he does not develop this principle other than to show that it could lead to his minimal state. He did not show why such a principle must lead to that choice. Had he started from his own principle, he would probably have made my work superfluous.

(*) “The Conceptual Foundations of Decision-making in a Democracy.”


THE EMERGENCE OF NORMS. Sugden’s “The Economics of Rights, Co-operation and Welfare” and part of Verbeek’s book devote some space to theories that attempt to explain the emergence of norms in terms of purely individual rationality. That produces ingenious and possibly useful results, but – as an undertaking that explains actual social decision-making – it is doomed to drown in its own paradoxes. These attempts are a good example of lack of the virtue of realism. They also neglect an essential variable of their model: the nature of the authority that imposes sanctions. The natural assumption would be that if this authority can enforce compliance to a norm, then it also will have the power to establish it. Even more serious is ignoring the fact that societies of pre-humans existed before norms. Like every society, they needed regulation and coordination of the behavior of its members in addition to the power of the leader. That function was performed by instinct. The norms that regulate a human society emerged as a means to deal with the man’s unique information processing abilities – reflective thinking, reason and imagination – which rendered instinct ineffective once one was aware of it. Norms result from the application of these same mental abilities to that problem. The fundamental, systematic distinction between the emergence of instincts versus that of our norms is that with animals the function of a feature emerges ex post, while the expected function of human products lies at the basis of their creation or at least of their acceptance, and is ex ante. History also shows that before democracy the prevailing authority indeed did control both the establishment and enforcement of norms. When dealing with norms, the first question then is: who has the authority to establish and enforce norms, and the second: which objective did he have in mind when doing so. A second objection I have to Sugden is his cavalier rejection of contract theory. As explained in 7d, he is correct in rejecting the notion of a real or imaginary ‘original situation’ as a basis for such a contract. But these are specific theories that do not invalidate the general concept of a social contract (see on this web-site “SOCIAL PHILSOPHY, JEAN HAMPTON”). As Verbeek does not rely on their emergence in his arguments in favor of his cooperative virtues, this does not invalidate his work. But, as said, the question of why people would acquire such dispositions and act accordingly, cannot be answered within the framework of a purely individualistic RCT.

How the first norms emerged in primitive societies will remain a mystery for which Sugden or Ken Binmore’s “Natural Justice” could provide a hypothetical explanation. But in historical times most norms were developed as rational responses of societies to problems whose numbers increased exponentially by technical progress. The priority rules of traffic laws are a ubiquitous example used in classic RCT’s. Here follows an imaginary, anecdotic explanation of know how the first ones came about.

The problem with traffic – at that time horse carriages – is the increase in numbers and quantity of vehicles and their speed (which is also influenced by the quality of the roads). Other factors increasing the urgency of the problem are the average haste of the people and their level of courtesy, or rather the lack of it (= collisions). If you know the history of Europe and have driven in both England and France, you will understand that the urgency was greatest in France, then under the reign of Louis the fourteenth. The occurrence that confronted the Sun King with the traffic problem was the collision between two young men (one the son of the king’s favorite mistress and the other the son of his best general) who careened with their carriages at full speed over a crossing of two well-maintained lanes in the forest ofr Fontainebleau. One was killed on the spot and the other died in the hospital before regaining consciousness. That is no way to die for the sons of such parents. They are supposed to die on the field of honor; the general’s son on the battlefield and the son of the mistress in one of the many duels he has to fight when called a bastard. Both parents were incensed, blamed the other’s son and asked the king for a verdict. Normally that would be a simple task, but on this case he held both in the same high esteem. History does not record how he extricated himself from that dilemma, but we do know that he decided to prevent it from recurring. He had an ample supply of highly educated civil servants and formed a commission. The first question, “do we need  traffic rules?”, was already answered in the affirmative. The next question was “right or left?” After months of rational deliberation in the spirit of Descartes, they came to a conclusion. “Most people are right-handed, so we should drive on the right. At crossings in cities and dense forests, the people coming from your side of the road, the right, have the shortest advance warning of our arrival and should get priority. So we should drive on the right and give priority to traffic coming from the right”. England, which scores the opposite in about all above-mentioned factors, experienced the necessity of priorities much later. When the question “which side?” came up in Queen Victoria’s cabinet, the deliberation was short. When told that the French drove on the right and gave priority to the right, the conclusion was immediate and unanimous: “If the bloody frogs drive on the right, we will drive on the left and give priority to the left”.  Given its lower rate of traffic accidents, the British decision must have been rational according to the classic definition of RCT.