Norms and Institutions: A Commentary on Aspects of N.K. Sundby's Philosophy of Law

Norms and Institutions: A Commentary on Aspects of N.K. Sundby's Philosophy of Law

Norms and Institutions:

A Commentary on Aspects of N.K. Sundby's Philosophy of Law

By Henning Herrestad

This paper was presented as Herrestad's "lecture of choice" at June 14, 1996, at the occasion of Herrestad's doctoral disputation at the Faculty of Law, at the University of Oslo.


Honourable chairman, honourable opponents, honourable professors, ladies and gentlemen,

I shall use this occasion to honour one of my predecessors - Nils Kristian Sundby. I shall do so by presenting Sundby's analysis of the concept of norms as a contrast to my own attempt to analyse this difficult concept. My overall aim is to present certain elements that I believe are central to the analysis of the social phenomenon known as "law". Hence, I shall do as philosophers of law always do. I shall discuss the concept of law.

Sundby has a unique position in the brief Norwegian history of the philosophy of law. In his review of Sundby's main work, Om normer (On norms), Knut Erik Tranøy hailed Sundby's book as "besides Torstein Eckhoff's book on justice, the first significant and modern norwegian work in the philosophy of law" (cf. Tranøy:76 p. 475). Sundby was a strong personality, an ambitious scholar, and a social reformer inspired by Jeremy Bentham's program of legal reform through critical assessment of basic legal concepts (cf. Andenæs:94 p. 253, Sundby:73 pp. 714-715). Sundby died in 1978, only 36 years old. He cannot, therefore, answer my criticism. I am a little anxious that some may feel that it is sacrilege to criticise Sundby's philosophy. I do not think it is. Many new contributions to the theory of norms have been published in the more than 20 years that have since passed. There is little reason to allow Sundby's analysis to remain as the one and only sanctified Norwegian philosophy of law. I am certain that Sundby would have agreed with this, were he still alive. However, I do not propose to give a fair and balanced assessment of all sides of Sundby's work. I shall only discuss two contentions I have about Sundby's theory.

First and foremost, I disagree with what I see as Sundby's debt to non-cognitivist moral and legal philosophy, his debt to emotivist ethics and Scandinavian legal realism. This debt is seen in his insistence that norms entail an inner compulsion to act in certain ways, a compulsion created through a process of internalisation. This debt is also seen in his insistence that sentences used to express norms cannot be true or false. I shall argue that this debt to non-cognitivism gives several problematic consequences.

Second, I disagree with Sundby's distinction between pliktnormer and kvalifikasjons-normer, which I translate as "norms of conduct" and "qualification norms". I shall maintain that this distinction confuses norms and the non-normative idea about a constitutive rule. However, I believe Sundby was right in giving emphasis to the role constitutive rules play in the analysis of law. At the end of this lecture, I shall follow Sundby's example by attempting to explain the role of such conventions for constitution of the legal system.

There are three other points of contention which I regard as important, but which I only have time to mention in passing. First, I disagree that Sundby's distinction between regler and retningslinjer (in English, "rules" and "guidelines") is a logical distinction. In my Masters thesis I argued that Sundby had a wrong conception of legal rules (cf. Herrestad:90 pp. 37-39). Svein Eng has made a similar point (cf. Eng:90 p. 480), and in the second edition of Rettssystemer, the book Torstein Eckhoff first wrote together with Sundby, Eckhoff accepts that there is no need for a logical distinction (cf. Eckhoff-Sundby:91 p. 111).

Second, I find unfortunate the comments Sundby made about deontic logic as being trivial, superficial and inadequate (p. 53). My impression that Sundby was misguided has been strengthened both from my own reading about deontic logic, and from Sundby's own attempts at formalisation. When Tranøy pointed out the logical consequences of his formal definition of permission, he revised it (p. 52, p. 64), in my view unsuccessfully. Sundby also gave a formal definition of the notion of competence (p. 414). Andrew Jones has pointed out its ambiguity and circularity. A detailed criticism is found in my doctoral thesis (cf. Herrestad:96 ch. 5.3).

Third, Sundby often showed a certain carelessness in his analysis. He introduces distinctions which he does not follow consistently. He sometimes expresses views that are ambiguous on the verge of being contradictory. I shall give an example. At his disputation the first opponent, Stig Strömholm, noticed how Sundby claims that he is doing conceptual analysis (cf. Strömholm:75 pp. 288-290). Certain claims about norms he dismisses as pointing to mere empirical contingencies irrelevant for the conceptual analysis. His criterion that norms must be internalised, on the other hand, is accepted even if it has a mere empirical justification. At again other places he dismisses arguments as mere logical questions. One can agree with Strömholm in wondering what criteria Sundby used for balancing conceptual and empirical issues.

Sundby's theory of norms

Let me try to give a brief account of Sundby's analysis of the concept of norms as found in part I of his thesis. Sundby starts his analysis by distinguishing different ways language is used in communication, different types of speech acts. Language may be used to describe reality. The speaker then asserts the truth of an indicative statement, conveying an indicative meaning content, and indicative intension, which gives a picture of how reality is. Language may be used to make other people change the reality. The speaker then expresses a normative statement conveying a normative intension. According to Sundby, there are two types of normative speech acts.

First, the speaker may express that a certain act shall be done. He may express that something is an obligation, a prohibition, or a permission. This type of speech act he terms a "directive". The existence of normative intensions conveyed by a directive speech act is the first constitutive element in what Sundby terms "pliktnormer". I shall term them "norms of conduct" or, when the context allows, simply "norms". The second constitutive element is that the normative intension is internalised in some person. I shall come to this a little later.

Second, the speaker may express that a certain expression or a certain state of affairs shall count as a certain other expression or state of affairs. For instance, the state of affairs that I write my signature shall count as the state of affairs that I give my consent. Sundby terms this type of speech act a "qualification". The existence of normative intensions conveyed by a qualification is the first constitutive element in what Sundby terms "kvalifikasjonsnormer",i.e., "qualification norms". Again, the second constitutive element is that the normative intension is internalised in some person. I shall discuss this category of norms in more detail later on. I shall first focus on the norms of conduct.

Sundby also mentions as separate speech acts the act of expressing certain feelings and the act of making a value judgement. This is important because it gives us reason to believe that he did not see directives simply as expressions of feelings, and he does distinguish between values and norms. Thus, he is not a pure emotivist in his theory of norms. However, Sundby does not give any clear criteria for distinguishing these various speech acts or the various intensions conveyed. His taxonomy is incomplete.

His main concern is to distinguish between normative and indicative intensions. He relies on the argument that these intensions differ with respect to their "direction of fit between the words and the world". He argues that only indicative intensions can fit the world; hence only indicative statements can be true or false. In my Masters thesis I give an argument as to why this is a misconception. In his doctoral thesis, Sundby has nothing more to say about the semantic properties of normative statements. In an earlier article written together with Ole R. Sandberg, Sundby suggests using the values gjeldende/ikke-gjeldende (cf. Sandberg-Sundby:71 p. 67). As Sundby resisted the association of these terms with the terms "valid/invalid", we may possibly translate them with the terms "holding/not holding", or "applicable/not applicable", or "in force/not in force". I prefer "holding/not holding". Sundby and Sandberg regarded normative statements as holding depending on whether the conveyed normative intensions were experienced as being binding, and on them being, by and large, obeyed. This suggestion ties the semantic evaluation of normative statements to whether or not the intension conveyed by the statement has been internalised.

Sundby maintains that norms are not only semantic categories, but psychological phenomena (pp. 17, 119). He maintains that a normative intension stands in the relation of being norm-for a certain person if and only if the person has internalised the normative intension. Hence, internalisation is a constitutive criterion for something being a norm. According to Sundby, the result of the psychological process of internalisation is that the person regards the normative intension as a duty for him or herself, or for others (p 125). He writes that the person in his rational judgement may disagree with what the norm prescribes as a duty, but that he feels bound to respect it. He asserts that the norm will give rise to a feeling of inner compulsion pressing the person in the direction of a certain course of action even though it is contrary to the persons strategic remonstrances. This, he notes, is the philosophically well-known feeling of duty (p 125). Though not all Sundby has to say about internalisation is equally clear, I interpret these statements as indicating that internalisation creates certain emotional dispositions, certain attitudes, which one cannot deny by rational argument. Hence, Sundby asserts that norms are non-cognitive. According to Sundby's semantic analysis, a normative intension is holding if and only if there exists such a compulsion, a feeling of being bound, in the mind of some person.

Strömholm regarded this psychological concept of norms as unfit for the analysis of law (cf. Strömholm:75 p. 288). However, Sundby also suggested a second way in which a normative intension may become a norm-for a person (pp. 149 f). The person may have internalised certain qualification norms which prescribe how norms may be enacted. When a normative intension is enacted in accordance with these qualification norms, the person will be disposed to internalise these normative intensions. Sundby maintains that, in virtue of this disposition, we may term the enacted normative intensions norms as well. Hence, not only normative intensions which have actually been internalised, but also normative intensions a person is disposed to internalise, qualify as holding and, hence, as being norms. By this argument, Sundby hopes to include all enacted law, as well as customary law, as being norms-for certain persons.

Given Sundby's insistence that internalisation is constitutive for something to be a norm, and his insistence that norms are psychological phenomena, a relation of a normative intension being a norm-for a certain person, I find it difficult to accept that enacted norms and internalised norms are treated as on a par with each other. It is like saying that everything written in my encyclopaedia is my beliefs because I am disposed to believe whatever I read there. In my view, my beliefs are restricted to information brought to my knowledge. The concept of belief cannot be extended to what I am disposed to believe. I may perhaps say that the information that I am disposed to believe is my potential beliefs. In the same way, I believe Sundby would have to term enacted norms potential norms as long as the potential for being internalized, the disposition, has not been realized. If Sundby really meant that internalisation is constitutive, then only normative intensions that actually have been internalised are real norms and not only potential norms. I can imagine, however, that Sundby would have disliked this conclusion, implicit in his theory, that most legal rules are only potential norms rather than real norms.

There are also several further reasons for regarding Sundby's internalisation criterion as problematic. As noted earlier, Strömholm points to a certain incoherence in putting so much emphasis on this empirically founded claim when other empirical claims are so easily dismissed. He questions whether the internalisation criterion is altogether necessary (cf. Strömholm:75 p. 290). I shall argue that, if Sundby had been more consistent about his non-cognitivism, he could have argued that internalisation was a necessary part of a conceptual argument concerning the link between norms and motivation. Tranøy, on the other hand, questions whether Sundby's internalisation criterion has committed him to certain unwanted consequences often associated with non-cognitivism (cf. Tranøy:77 p. 52 and 59). On Sundby's account, internalisation is a kind of psychological conditioning process establishing a non-cognitive compulsion to act in a certain way. This appears to rob Sundby of the possibility of arguing that norms and compliance to norms can be rationally justified. This is an issue which deserves some further elucidation. Hence, I shall now describe some of the philosophical background for Tranøy's warning.

Sundby's debt to non-cognitivist moral and legal philosophy

When I earlier referred to Sundby's debt to non-cognitivist moral and legal philosophy, I was primarily thinking about the internalisation criterion and his denial that normative statements can be true or false. The founder of Scandinavian legal realism, Axel Hägerström, for instance, explained people's obedience to norms as the effect of social training by parents, teachers and other authority figures, a training which had the result of establishing an inner compulsion to act as they were commanded (cf. Schmidt:78 p. 157). Hägerström's pupil, Carl Olivecrona, writes that normative intensions are conveyed by a special sort of performative speech act. Following John Austin, he regards normative statements as incapable of being either true or false (cf. Schmidt:78 p. 171).

A quick look at Sundby's references will confirm that another of his sources of inspiration was emotivist moral philosophy (pp. 416--433 ). As noted, he was not a thorough emotivist, but he shared some of its fundamental assumptions. I shall now argue that if Sundby had read his sources more attentively, emotivist moral philosophy might have given Sundby a better justification both for denying normative statements truth-values and for holding the internalisation criterion to be conceptually important. Realist moral philosophy, on the other hand, has provided me with my cognitivist view of norms, which I will describe later. Hence, I shall briefly say a few words about moral realism as well. In this account of moral philosophies, reference is usually made to moral values. Moral norms may either be thought of as existing on par with moral values or as being justified by these moral values.

The main source of inspiration of emotivist moral philosophy is the moral philosophy of David Hume. In moral philosophy every new philosopher has to take a stand for or against the philosophical arguments of David Hume. As most of you will know, Hume argued against a moral realism which regarded moral properties as residing in the external world (cf. Hume:84 pp. 507-527). He argued that the external world was material only, devoid of moral qualities, and that any philosopher who deduced a moral conclusion about what ought to be the case from merely factual premises concerning what is the case was making a logical error.

Hume also defended a particular philosophy of mind, based on a hydraulic metaphor, in which the cognitive faculties are the levers directing the kinetic energy of the water, and where the moving energy, the water, is irrational sentiment. What moves us into action are our likes and dislikes. However, the rational itself cannot motivate; it can only provide a means-end calculation about how best to achieve what we like. We can only reason about how existing desires or ends can best be satisfied. We cannot use reason to change these desires or ends. They are causally, not rationally, effected. Hume identified moral values with these irrational desires; hence moral values cannot be established by reason.

Hume's ideas are still very influential. A.J. Ayer followed Hume's ideas to their logical consequence. He has been recorded as saying that the whole of ethics could be written on the back of a postcard (cf. McNaughton:88 p. 17). As morals are based on irrational feelings, to say that stealing is wrong is simply to ventilate one's negative attitude to stealing. Instead of saying that stealing is wrong, one might as well have said "Boo!". Instead of saying that caring for the sick is good, one might as well have said "Hurrah!". Not surprisingly, this view was dubbed the Boo-Hurrah theory of ethics.

As facts may be ascertained by observation and experiment, our beliefs are subject to revision, but there is an external factual reality to which our beliefs must conform. If our beliefs fit the facts, they are true; if not, they are false. Not so with moral values. In the emotivist view, there are no moral facts about which to hold true or false beliefs. Instead, moral evaluations are affective responses to the facts of the world. They cannot be true or false. I believe these ideas were behind Sundby's claim that normative statements do not have truth-values.

According to emotivism, science attempts to justify in a rational way which of our beliefs are true and which are false. As feelings are non-cognitive, they cannot be rationally justified. Hence, if moral values are based on feelings, they are like one's tastes. Some like cold rice pudding, some don't; some like to cause pain, some don't. If sentences conveying moral arguments cannot be true or false, then moral arguments are unfit to be either premises or conclusions in logical inferences. Where moral talk dresses up as logical arguments with a cognitive content, this is only dressing in borrowed clothes, a mere rhetorical trick in order to gain acceptance. Moral values and norms cannot be part of any rational argument. To change the attitude of the person who likes to cause pains is not an intellectual task, but a matter of causing him to feel otherwise. Hence, there is no room for moral justification, only for moral persuasion. This view is very well illustrated in the film A Clockwork Orange, where the main character, who has the anti-social attitude of enjoying causing pain to others, is sentenced to a psychological treatment conditioning his personality to abhor violence.

This is the simplest version of emotivist ethics. The appeal of this theory is mainly its absence of ontological commitments to moral values as non-natural entities existing alongside the material world, an absence of epistemological commitments to a sixth sense - a moral intuition - by which these non-natural entities are experienced. Ideologically emotivism had a radical direction against the assertions of various conservative and religious groups that their moral views are based on facts.

Moreover, emotivism gives a suggestive explanation of the action-guiding character of morality. If you can cause a person to have a negative attitude towards causing pain, then you have caused the person to be motivated not to cause pain, and to act disapprovingly of others who cause pain. If the feelings are providing the motivational push behind an action, then to say that moral values are attitudes is to say that there is a direct conceptual link between moral values and motives for action. The authority of moral demands is seen as stemming from the fact that moral demands are anchored in a person's interests and desires. I believe Sundby proposed his internalisation criterion in order to explain the action-guiding capability of norms.

If norms are attitudes, then internalisation is just the creation of an attitude. Hence, emotivism makes sense of Sundby's internalisation criterion as a conceptual criterion, not merely an empirical association.

In its radical form, emotivist ethics faces several problems. There is something very disturbing and strongly counter-intuitive about a theory stating that the whole of morality is beyond rational consideration, that it is non-cognitive. It is disturbing because all values and norms appear to be arbitrary; any value appears to be as good as any other. The only difference between values is that some people are in a position where they have the power to influence other people into accepting certain values as superior. There are no boundaries to what may some day be held to be morally acceptable. There are no ways to reach genuine rational agreement on what is right or wrong, what is good or bad. I believe it is to these disturbing consequences of non-cognitivism that Tranøy is pointing in his review of Sundby. Moreover, I believe that such a view of the individual as being socially conditioned to act in certain ways leads to a social theory which does not give enough emphasis to the principle of rational choice. This is the regulatory idea that whenever one attempts to explain social phenomena, such as people's adherence to law, one should first attempt to find a rational explanation before concluding that the observed behaviour is irrational. I shall argue that people can also be seen to follow norms because they believe compliance is rationally justified.

Emotivist ethics can also be criticised for being counter-intuitive. Common intuition tells us that we can in fact argue about what things are right or wrong, good or bad, in just the same way as we can argue about the facts concerning, let's say, "black holes" or "the optimal design of a bicycle". People appear to reason about their values and norms, and people appear to require certain values and norms to be rationally justified. People appear to be able to reject values and norms when arguments show them to be inappropriate, and people appear able to accept new values and norms because they are rationally justified.

Much of contemporary moral philosophy tries either to mend emotivism to make it less disturbing and less counter-intuitive, or to defend the case that emotivism is wrong, arguing that we should accept moral realism after all. Richard M. Hare mends emotivism by acknowledging the cognitive element in norms while retaining the emotive element as fundamental (cf. Hare:61, Hare:63). If a more detailed comparison should be made between Sundby's theory of norms and moral philosophy, I suggest Hare's theory as a good place to start. I believe futher comparison between theories about morals and theories about law may yield valuable insights to both fields, but this issue will not be treated here. At the moment I shall only say a few words about moral realism, as this type of ethical theory has inspired some of my own views about norms and values.

I am still not entirely convinced about the central tenet of moral realism, the tenet that moral facts exist. Moral realists, like David Brink or David McNaughton, argue that moral facts can be regarded as supervening on natural facts, just as we regard mental facts as being supervening on natural facts about the brain (cf. Brink:89, McNaughton:88). Brink argues that Hume's is-ought gap is no more threatening than our inability to explain how mental facts actually connect to electro-chemical states in the brain (cf. Brink:89 ch. 6). He argues that reasoning from what is to what ought to be is no more problematic than to reason from natural facts to mental states. Our beliefs about moral facts are still inherently corrigible, as we have no better way to justify these beliefs than factual beliefs.

Gilbert Harman argues, against moral realism, that to maintain that moral facts exist has no explanatory value. Hence, we can use Ockham's razor on the tenet (cf. Harman:77 p. 7). On the other hand, the same argument can also be used to exclude existence of natural facts, as attested by the ongoing realist/anti-realist debate in the philosophy of science. I believe moral realism, and also theories about natural law, ought to take into account more of this debate in the philosophy of science, and also the current debate about mental states in cognitive science and artificial intelligence.

Moral realism defends a cognitivist view of norms and values. If there are moral facts, then a person's moral values and norms are simply beliefs about these facts, and moral statements are simply descriptions of these facts. Moral descriptions will be either true or false according to whether or not they correspond to the actual moral facts. Then there are no more problems about making logical deduction. We can then argue rationally about which moral values and norms can be justified as being true, just as we can argue rationally about which other beliefs about the world can be justified as being true. As implied by Harman's argument, however, we do not need to assume moral facts in order to defend a cognitivist view of norms and values. After all, we also accept, and try to justify rationally, many different beliefs about norms and values that everybody accepts to be conventional. Hence, we do accept that beliefs about norms and values can be justified even when we do not assume these beliefs to be beliefs about external facts supervening on natural facts. We may also accept that individuals hold beliefs about norms and values which are not conventionally shared. I shall maintain that statements about such norms and values are true or false depending on whether or not the corresponding personal or conventionally established beliefs exist.

This is not to say that all actual norms and values can be rationally justified. Jon Elster argues that all general attempts to reduce all behaviour based on social norms to some type of optimising behaviour have failed (cf. Elster:89 p. 115). Though reduction is often possible, it is not always possible. I just want to maintain that it is possible, in principle, to justify certain norms and values, and to discard irrational norms and values when better justified norms and values are found, just as we can change our other beliefs through rational argument. This is my cognitivism about norms and values. I believe this view to be more intuitively plausible than Sundby's non-cognitivist theory, for which a rational justification and revision of norms and values seems out of reach.

There remains a problem about how such rationally justified beliefs can be motivational. McNaughton maintains the tenet of an internalism between moral values and motivation, but rejects Hume's view that rational beliefs cannot motivate by themselves (cf. McNaughton:88 pp. 106-115). Brink instead rejects internalism (cf. Brink:89 pp. 45-50). He accepts that there is no direct link between holding a certain belief about what is moral and being motivated to act in this way. He argues that moral norms are followed because of our ability to rationally justify those moral norms as being coherent with our other values and desires. The ultimate justification will appeal to the reflective equilibrium of our various desires and valuations (Brink:89 p. 66). According to Brink's analysis, the problem with Hume's theory is not his view that values and desires are motivational, but his view that values and desires are beyond rational criticism. Both of these accounts allow us to reject Sundby's internalisation criterion without losing the ability to explain how norms and values can be motivational.

Before I go on to discuss my second point of criticism with regard to Sundby's theory of norms, I feel a need to say just a few words about the distinction between norms and values. Like Sundby, I believe the content of a norm to be a normative intension stating that some action is obligatory, prohibited or permitted. I believe the content of a value to be a normative intension stating that something "ought to be the case". Such beliefs are sometimes termed "value judgements", sometimes "wishes", sometimes "preferences". There may be reasons to distinguish these concepts as different modalities, and there may also be several other modes of ideality - to paraphrase Wertheimer. However, I do not think this is necessary for the present purposes. Sundby suggests that in legal argumentation inferences are made from values to norms and vice versa, but he does not suggest any conceptual connections. G.E. Moore suggested that there was a conceptual connection making it obligatory to bring about the state of affairs that ought to be the case (cf. Moore:03 p. 147). Sven Ove Hansson argues that this connection yields counter-intuitive consequences which are avoided by making the opposite connection; he argues for a the existence of a conceptual connection such that it is prohibited to bring about a state of affairs that ought not to be the case (cf. Hansson:91 p. 10). If one holds certain states of affairs to be of neutral value, the two connections are not identical. This issue about the possible interrelations between norms and values is yet another issue which merrits further investigations.

Conventions and constitutive rules

As mentioned earlier, Sundby distinguished between norms of conduct, containing the modal character that some action shall be done, and a special category of norms termed "qualification norms", having the modal character that something shall count as something else (p. 78). He admits that his term "norm" covers more than the conventional usage of the term (p. 77). Norms are usually thought to be action-guiding, while qualification norms are not, but Sundby does not consider this to be decisive against his classification. Apparently one of his reasons for including the intension of a speech act of type qualification in the category of norms is his assertion that both are conventional rather than natural phenomena (p. 78). Both our norms of conduct and our qualification norms could have been otherwise, and they do change from place to place and from time to time. I believe that in making this comparison, Sundby mistakenly identifies the conventional and the normative.

As pointed out by Andrew J.I. Jones and Ingmar Pörn, it is constitutive of the category of norms that they can be violated (cf. Jones-Pörn:85 p. 279). For instance, there is a rule of conduct in football games which demands that the referee record a goal each time the ball crosses the goal-line. This rule also allows violation. The referee may pretend he did not see it. On the other hand, the rule of football qualifying the state of affairs that the ball crosses the goal-line as "a goal", is a rule that cannot be violated. The fact that the ball crosses the goal-line counts as a goal in football no matter how seldom goals are actually recorded. We tend to say that it counts as a goal by definition. You cannot violate a definition. Such definitions, or definition-like rules, define what football is; they constitute the game. Hence, they are termed "constitutive rules". They are typical examples of what Sundby termed qualification norms. My point is this; as constitutive rules cannot be violated, they do not share the most fundamental characteristic of norms. But, of course, both the rules of football that express such constitutive rules and those expressing norms of conduct are conventional and could have been otherwise.

Sundby also maintains that constitutive rules are norms because they satisfy his two criteria for something being a norm-for a person. He maintains that they are intensions that are normative, and that they are internalised. I have just argued that they cannot be seen as intensions that are normative. Sundby maintained that these intensions have a modal character stating that something shall count as something else. It is less controversial, however, simply to describe this character without the shall; simply as the character that something counts as something else. This is more in accordance with John Searle's taxonomy of speech acts (cf. Searle:79 pp. 26-27), and it is in accordance with the recent analysis by Andrew Jones and Marek Sergot, where constitutive rules are regarded as a particular institution-relative conditional termed a "counts as" conditional (cf. Jones-Sergot:95).

I have already argued against the internalisation criterion as a constitutive criterion for norms of conduct. I find the internalisation criterion even less adequate for distinguishing constitutive rules as a type of norms. Sundby makes a rather loose reference to Noam Chomsky, maintaining that linguistic constitutive rules are internalised in language users (p. 143). What Chomsky means by this is not clear, however. I doubt that Chomsky means that language users feel an inner compulsion to follow grammatical rules even when they find it rational not to do so. In the same way, I can possibly imagine a football referee who feels an inner compulsion to register a goal, even if he really favours the other team and wishes not to. But I cannot imagine a football referee who feels an inner compulsion to regard the fact that the ball crosses the goal-line as being a goal, even if he wishes not to. Hence, I have problems understanding how the internalisation criterion makes sense when applied to constitutive rules. Strömholm asked whether it is in fact possible to combine the internalisation criterion with the distinction between qualification norms and norms of conduct (cf. Strömholm:75 p. 290). He may have shared my problems.

My main contention with the internalisation criterion is that it appears to preclude an analysis of institutions such as law which respect the regulatory idea of rational choice. When we ask why people accept certain conventionally established constitutive rules, an answer has to be found concerning how these people are socially conditioned to accept these rules, rather than to answer why people choose to accept these rules because they find them rationally justified. I do not maintain that all beliefs about constitutive rules can be rationally justified, but I do believe it to be a valuable methodological principle to attempt to give a rational explanation before an explanation about irrationality is accepted. I believe Sundby may have agreed to this in principle, but that he was unable to rid himself of his non-cognitivist presuppositions.

Strömholm's suggestion was that Sundby ought to have seen the legal system as primary, with respect to individual legal norms, and that this view can be generalised to all types of norms (cf. Strömholm:75 p. 290). This view that systems of norms are irreducible entities has several adherents. At least it is my impression that this is a tenet of the institutional theory of law of David N. McCormick and Ota Weinberger (cf. Lagerspetz:89 p 13). I do not believe these philosophers think of legal systems as having an independent existence such as the moral facts of moral realism. Rather, I suppose they have just chosen the existence of legal systems as a theoretical first principle. None the less, such a theory is in conflict with the regulatory idea in social science that the only primitives are individual mental states, individual beliefs in particular, but also individual beliefs about other people's beliefs, and their mutual beliefs about each other. This is a form of methodological individualism, an attempt to avoid hypotheses about supra-individual social forces.

Is it possible to give an explanation of social institutions such as law which conforms to the regulatory ideas of rational choice and methodological individualism? I believe the first step towards such an analysis was taken by David K. Lewis in his book Conventions from 1969. He suggests that linguistic conventions may be seen as co-ordination equilibria in a certain game-theoretical set-up. This rational basis for conventions need not be consciously held by all participants, but it is revealed in cases where the co-ordination equilibrium is disturbed. A new consensus must then be negotiated. A further step was taken by the Finnish philosopher Eerik Lagerspetz in his doctoral thesis A Conventionalist Theory of Institutions from 1989. Lagerspetz has an interesting theory about how conventions are reducible to mutual beliefs in a group of people. Lagerspetz suggests various game-theoretical set-ups which may explain the rationality of such mutual beliefs which form the basis of social phenomena such as the acceptance of political authorities, the acceptance of money, and the role of threats in the legal system. This division of the law into a series of different sub-systems, which are seen as rational in the light of various game-theoretical set-ups, is still another promising field of research.

Lagerspetz's idea that conventions are reducible to mutual beliefs in a group of people can explain how conventional norms may have both a factual and a non-factual aspect. Consider that in a certain society a convention exists about a constitutive rule stating that "if the sovereign speaks ex officio, his words count as legal rules". The result of this convention is that legal rules get established as what John Searle termed "institutional facts" (cf. Searle:69 p. 51). Institutional facts have a non-factual character in that they are nothing more than beliefs shared by a group of individuals. The existence of the mentioned sovereign's competence or power to enact legal norms still has no other ultimate basis than a shared belief among a sufficient number of people that this competence exists. In this respect, legal competence has no more reality than mermaids and unicorns. In this sense, the philosopher Axel Hägerström was correct that an institution like the legal system is based on superstition (cf. Schmidt:78 p. 156). However, as argued by Lagerspetz, it may be entirely rational to have such beliefs.

Institutional facts also have a factual character in that no single individual can meaningfully deny the existence of the institutional facts of the group in which he lives. You can meaningfully deny the existence of mermaids, but you cannot meaningfully deny the existence of, for instance, money; you can meaningfully deny the existence of a moral obligation to comply with a certain properly enacted legal requirement, but you cannot meaningfully deny the existence of the legal requirement itself. Despite their mere notional existence, they are beliefs that are important points of co-ordination for the members of our society. Lagerspetz emphasises their importance for the co-ordination of actions (cf. Lagerspetz:89 p. 72-73). I believe that they can probably be seen as important co-ordination points in our cognitive maps of our social world as well.

This idea that conventions are reducible to mutual beliefs ties in with my theory of norms and values as cognitive entities. It explains how enacted norms and values, on the one hand, are reducible to beliefs held by certain individuals, and, on the other hand, how their existence can be ascertained independent of whether they are actually believed by those who enact them or by those who are subject to them. To paraphrase H.L.A. Hart, the whole legal system may be seen as resting on certain mutual beliefs among a significant number of legal officials about certain constitutive rules concerning what counts as legal rules in the system (cf. Hart:61 p. 89-114). If the legal officials have the social convention that laws enacted by parliament count as legal rules, then, on the one hand, law is reducible to these shared beliefs held by a group of individuals, and on the other hand, what the parliament enacts now becomes law regardless of the beliefs of the individual members of the parliament and regardless of the general acceptance of these beliefs in the rest of society.

In this way, both norms, values and constitutive rules may exist as institutional facts which may be ascertained to exist in two ways: either directly, being mutually believed by a group of people, or indirectly, because a group of people has the mutual belief in a constitutive rule about what counts as norms, values and other constitutive rules. Moreover, we are now able to explain how a normative statement may have truth-values even when it is not a description of existing beliefs, nor of the current beliefs of the speaker or listener. We may regard the speech act in which a normative intension is being expressed as a process of enactment. In other words, the normative statement can be regarded as a performative which creates a new norm at the moment of being expressed (cf. Jones:83 pp. 94-97, Lewis:83 pp. 220-226, Sesonske:65). At the same time the normative statement is descriptive of the norm it is creating. The statement is true if and only if the norm has been created; it is false otherwise.

Why should the rest of society accept that what the legal officials conventionally believe to be law is law? Hart's theory has a problem in that the rule of recognition is not itself thought to be a legal rule, but a social convention among the legal officials. However, the rule of recognition is also thought to identify rules of adjudication, which in turn state who the legal officials are. But the establishment of a social convention seems to presuppose that the group of legal officials existed before the convention. Hence, the convention cannot itself establish the group. Lagerspetz breaks this circle by insisting that the initial group of legal officials are not identified by a pre-existing rule, and that a political division of labor, by which people accept some persons as authorities, is more basic to the legal system than Hart's rule of recognition (cf. Lagerspetz:89 pp. 117-119). First there appeared a social convention in society at large identifying certain persons as legal authorities. Then came the convention among these authorities about a rule of recognition for identifying legal rules and for identifying their equals and successors. Why do people accept a political division of labour in the first place? Lagerspetz explains this acceptance of authorities as a rational solution in co-ordination situations where the cost of bargaining is too high. However, it is rational to accept certain authorities only as long as they are relatively impartial and effective in providing co-ordination.

This gives certain limits to the legitimacy of the rule of law. In a legal system where the legal authorities are so inequitable or so ineffective that it is no longer rational to accept their authority, the system may collapse. Another limiting factor is that a convention will only exist as long as there exists a mutual belief about it among, what Lagerspetz terms "the relevant persons" (cf. Lagerspetz:89 p. 21). Consider the case of a constitutional meeting of a new association where they first decide that constitutional issues must be decided by a two-thirds majority vote. In making this decision, all the persons assembled are equally relevant; thereafter the relevant persons equals any two-thirds majority of the assembly. In the case of complex social institutions such as a legal system, I believe our model is one of concentric circles of people with less and less influence on what institutional facts are created or rejected. At the centre, each person's relevance to the system is so weighty that what is mutually believed by only a few persons is of significance. As the fate of the former East Germany illustrates, the mutual beliefs of the persons at the periphery also have a certain relevance when their numbers become sufficiently large.


I have attempted to draw a picture of the social institutions of law which is in accordance with the regulatory ideas of methodological individualism and rational choice. In summary, the law is based on certain conventions, which are in themselves nothing more than mutual beliefs among a significant number of persons relevant to the law. If the conventions are about certain constitutive rules, norms and values and still other constitutive rules are established as institutional facts. All of these facts can be said to have a mere notional existence. In this I agree with Scandinavian legal realism. But I believe that these institutional facts can also be rationally justified by their function as important points of co-ordination for both social interaction and cognitive orientation.

I agree with Sundby that the content of norms can be described as equal to the intensions of a directive speech act. However, these normative intensions stand in the relation of being norms-for a person simply by being believed by that person. I have argued against Sundby's criterion that the intension must be internalised in the person such that he or she feels an inner compulsion to act in certain ways even when he or she rationally disagrees. In my view, both norms and values are cognitive elements. Furthermore, contrary to Sundby, I argue that statements concerning values and norms have truth-values; they either describe a belief about a normative intension or they describe a norm enacted by uttering the statement itself. If normative statements are descriptive and have truth-values, it is unproblematic to explain how they can be part of rational argument and logical deduction. I have also argued that compliance with norms and adherence to values can be motivated by giving a rational justification for it.

Contrary to Sundby, I have argued that the basis for establishing values and norms by enactment are conventions concerning non-normative constitutive rules rather than the special kind of norms Sundby termed "qualification norms". I see them as non-normative institutional facts because they have the character of being definitions, and, as such, they cannot be violated. Sundby was right, however, about their fundamental importance to the creation and maintenance of institutions such as law. It is a limiting factor to an institution such as law, that in the end it is based on certain beliefs and the conviction that these beliefs are rational. If a significant number of people cease to have these beliefs, the institution collapses.

I may sound overly rationalistic in my theory about norms and institutions. I must again emphasise that I am a long way from believing that people only hold beliefs, or act on beliefs they are able to rationally justify. Neither do I believe all conventions and institutions are rationally justified. But I believe in the idea that when one attempts to explain social phenomena, one should first attempt to find a rational explanation before concluding that the observed behaviour is irrational. There are of course social phenomena which are difficult to explain as the outcome of rational decisions. What I find both disturbing and counter-intuitive, however, is a theory of norms, values and conventions which pre-empts the possibility of seeking any rational explanation of social phenomena. I hope to have mended Sundby's theory of norms in this respect.


Kristian Andenæs, "Kritisk juss: Hva nå?", Kritisk Juss, 4:247--266, 1994.

J.L. Austin, How to Do Things with Words, Oxford University Press, Oxford, 1962.

David O. Brink, Moral Realism and the Foundations of Ethics, Cambridge University Press, Cambridge, 1989.

Jon Elster, "Social Norms and Economic Theory", Journal of Economic Perspectives, 4(4):99--117, 1989.

Svein Eng, "Sondringen mellom regler og retningslinjer", Festskrift til Anders Bratholm, pp. 473--483, Oslo, 1990.

Torstein Eckhoff and Nils Kristian Sundby, Rettssystemer - Systematisk innføring i rettsfilosofien, 1. utgave, TANO publishers, Oslo, 1976.

Torstein Eckhoff and Nils Kristian Sundby, Rettssystemer - Systematisk innføring i rettsfilosofien, 2. utgave, TANO publishers, Oslo, 1991.

Sven Ove Hansson, "Norms and Values", Crítica, 23 (67):3--13, 1991.

R.M. Hare, The Language of Morals, Clarendon Press, Oxford, 1961.

R.M. Hare, Freedom and Reason, Oxford University Press, Oxford, 1963.

Herbert L.A. Hart, The Concept of Law, Clarendon Press, Oxford, 1961.

Henning Herrestad, Norms and Formalization, CompLex 12/90, TANO publishers, Oslo, 1990.

Henning Herrestad, Formal Theories of Rights, Juristforbundets Forlag AS, Oslo, 1996.

David Hume, A Treatise of Human Nature, (first published 1739), Penguin Classics, London, 1984.

Andrew J.I. Jones, Communication and Meaning - An Essay in Applied Modal Logic,

D. Reidel Publishing Company, Dordrecht, Holland, 1983.

Andrew J.I. Jones and Ingmar Pörn, "Ideality, Sub-Ideality and Deontic Logic", Synthese, 65(275--290), 1985.

Andrew J.I. Jones and Marek Sergot, "A Formal Characterisation of Institutional Power", Prepublished as part of proceedings from MEDLAR II workshop in Rome (August 1994), 1995.

Eerik Lagerspetz, A Conventionalist Theory of Institutions, Acta Philosophica Fennica, vol. 44, Helsinki, 1989.

David K. Lewis, Convention, Harvard University Press, Cambridge, Massachusetts, 1969.

David K. Lewis, "General Semantics", in David K. Lewis, editor, Philosophical Papers,

vol. 1, pages 189--232. Oxford University Press, Oxford, 1983.

David McNaughton, Moral Vision - An Introduction to Ethics, Blackwell Publishers, Oxford, 1988.

George Edward Moore, Principia Ethica, Cambridge University Press, Cambridge, 1903.

Ole Rømer Sandberg and Nils Kristian Sundby, "Direktiver og logikk", Tidsskrift for rettsvitenskap, 83(2):17--68, 1971.

John R. Searle, Speech acts - An essay in the philosophy of language, Cambridge University Press, Cambridge, 1969.

John R. Searle, "A Taxonomy of Illocutionary Acts", in John R. Searle, editor, Expression and Meaning, pages 1--29, Cambridge University Press, Cambridge, 1979.

Alexander Sesonske, "Performatives", The Journal of Philosophy, 62:459--468, 1965.

Stig Strömholm, "N.K. Sundby: Om normer", review in Svensk Jurist Tidning, 285--291, 1975.

Nils Kristian Sundby, "Benthams betydning for vår tids rettstenkning", Tidsskrift for Rettsvitenskap, 85:676--722, 1973.

Nils Kristian Sundby, Om Normer, Universitetsforlaget, Oslo, 1974.

Knut Erik Tranøy, "Nils Kristians Sundby: Om normer", review in Lov og Rett, 10:475--479, 1976.

Knut Erik Tranøy, "Sundby - Om normer", review in Norsk Filosofisk Tidsskrift, 1:47--60, 1977.

Publisert 17. des. 2009 19:33