Value judgments abound in law and legal science. Many legal rules explicitly mention values in their conditions of applicability. For instance, Article III-I: 103 Section 1 of the Draft Common Frame of Reference states that ‘a person has the duty to act in accordance with good faith and fair dealing in performing an obligation …’. Article 4:101 of the Principles of European Tort Law reads that ‘A person is liable on the basis of fault for intentional or negligent violation of the required standard of conduct’. Moreover, according to Article 3:201 of the same Principles, the attribution of damages to the tort-feasor depends on ‘… the nature and the value of the protected interest; …’.
Even where legal rules do not explicitly invoke valuation, legal decision makers attempt to give decisions which are not only in accordance with the (positive) law, but which are also equitable. Therefore legal decision making also requires value judgments.
It has been argued that legal science should concern itself with the normative question what the law ought to be or with which norms ought to be enforced by collective means. Strictly speaking these ought questions do not require value judgments for their answers because ought judgments are not value judgments, but in practice ought judgments are often based on value judgments, and therefore these two views on the (main) purpose of legal science assign a central role to value judgments.
In this book one particular evaluative question takes a central place. It is the question on which level the private law in Europe is best created. Many different agents present themselves as potential creators of law, both private and public agents, on levels varying from municipalities through countries and the European Union (EU) to the whole world. In particular within the EU the issue whether regulation should take place on the national or on the EU level is hotly debated.
This assigns the question how value judgments can be justified a high ranking on the list of questions that need to be addressed by legal theory in general and by theoretical reflection on the foundations of European private law in particular. And yet the question has not received a lot of attention. There are many books on legal decision making, interpretation in law, and legal justification but they seldom address the question how value judgments, rather than legal judgments in general, are to be justified.
It is the purpose of the present chapter to remedy this shortcoming of legal theory, at least to some extent. A full discussion of the justification of value judgments requires lengthy excursions into the fields of ontology, epistemology, practical reason, and ethical theory. Here I will try to sidestep these complications as much as possible, while recognizing that this leaves important issues unexplored.
The argument of this article is divided into two main parts. The first part deals with justification as if it were an argument that leads from a set of premises to the value judgment as its conclusion. Within it, a subdivision is made between approaches to justification that take deductively valid arguments as their model (the ‘deductive’ approach), and approaches that are based on the balancing of reasons for and against the value judgment (or any other conclusion). The deductive approach (sections 3-7) will be used to discuss the main issues, and the reason-based approach (section 8) is sketched as a refinement of the deductive approach.
In the context of the deductive approach to justification, three issues that are of general relevance for the justification of value judgments receive special attention. One of these issues concerns the supervenient nature of value judgements, and the related topics that value judgments can be universalized and that they presuppose standards which are not true or false. These basic topics are discussed in the sections 4.4 and 4.5.
A second issue is whether an attempt is made to find a foundation for value judgments in premises which are beyond dispute, or whether such an attempt is not made. The former approach to justification is called ‘foundationalist’, while the latter approach is called ‘coherentist’. The foundationalist approach is discussed in the sections 5 and 6, while coherentism is discussed in section 7.
A third issue concerns the distinction between ‘ordinary’ value judgments, such as the judgment that a particular law is just, or that an agent acted unlawfully, and comparative judgments, such as the judgment that contract law is better regulated on the level of the EU than on state level. Comparative judgements are discussed in section 9, after the reason-based approach.
The second part of the article (section 10) deals with justification as a kind of dialogue between the person who pleads for the value judgment, and an audience that is opposed to it, or at least needs to be convinced of it. This distinction between these two approaches is basically one between a static timeless approach to justification and a dynamic or procedural one, according to which justification is a process that stretches out in time.
The article is summarized and some conclusions are drawn in section 11. In section 12 an attempt is made to translate the findings of the previous sections into a method for justifying value judgments in general, and value judgments about the best level of regulation for European private law in particular.
Justification can very well be seen as a kind of argument, with the judgement that is to be justified as its conclusion. It is often assumed that such justificatory arguments are deductive ones, and more in particular syllogisms. The idea of syllogistic justification is quite simple: a judgment is justified by a syllogistic argument with two premises. The one premise formulates a kind of law, rule or standard. It has the form of a conditional sentence: ‘If…. Then …’. The other premise contains facts about a particular case, situation, or kind of action, facts that match the condition (if-part) of the first premise. This syllogistic model of justification can be used for analysing the justification of value judgments, but also for the justification of purely factual judgments, predictions and normative (must, should, ought) judgments.
Let us have a look at some examples. The first one concerns the justification of an ordinary ‘factual’ judgment:
If two witnesses testified that Jerry murdered his mother in law, then Jerry did commit that murder.
Both Jerry’s wife and his father in law testified that they saw Jerry murder his mother in law.
Judgment that is being justified:
Jerry murdered his mother in law.
The second concerns the justification of a prediction:
If the interest rates rise, the stock market will go down.
The interest rates are climbing.
Judgment that is being justified:
The prices of stocks will decrease.
The third is the justification of a should-judgment.
If somebody promised to do something, then she should do it.
Jane promised to buy an ice cream for Peter.
Judgment that is being justified:
Jane should buy an ice cream for Peter.
And finally an example about a value judgment:
If an audio set satisfies the DIN 45500 standard, it is good.
Jan’s audio set satisfies the DIN 45500 standard.
Judgment that is being justified:
Jan’s audio set is good.
This syllogistic model of justification has several attractions, not the least of which is that there is one uniform model that covers very heterogeneous kinds of justification. However, it is not without complications, and these complications become very visible if the syllogistic model is applied to the justification of value judgments. Issues that arise have to do with the correctness (‘truth’) of both the standard that is being applied and of the value judgment that is being justified. For a proper understanding of these complications, it is necessary to take a closer look at the nature of value judgments.
The justification of value judgments, such as ‘This is a good soccer match’ is different from the justification of other judgments, such as the judgment ‘There is a horse right in front of me.’ To understand properly why this is the case, it is useful to have a closer look at the nature of valuation. This section addresses some of the relevant issues, in particular the naturalistic fallacy, evaluative non-cognitivism and constructivism as views of what is involved in giving and justifying value judgments, and the supervenience and universalizability of value judgments as essential preconditions for the justification of value judgments.
Originally the naturalist fallacy had to do with attempts to define evaluative terms, and in particular evaluative terms as used in moral judgments. For a proper understanding of what the naturalistic fallacy amounts to, this connection to morality is not crucial, however. In the following discussion, the examples will be taken from law and sport.
A bachelor is an unmarried man of marriageable age. This is not an empirical finding; it is a matter of definition. The word ‘bachelor’ just means ‘unmarried man of marriageable age’. This definition of ‘bachelor’ specifies both a necessary and sufficient condition for the applicability of the word ‘bachelor’ and in that way gives the meaning of the word.
Suppose that we have a similar necessary and sufficient condition for the applicability of the expression ‘good faith’ in contract law, would we then also have the meaning of this expression? Lawyers sometimes create the impression that they believe so, when they write that they want to know what ‘good faith’ means in connection with this or that statutory provision. But are they right when they formulate their interest in when a contract party acts in good faith in this way?
At the beginning of the 20th century, the Cambridge philosopher Moore offered a very influential argument why this would not be the case. He considered the possibility that we had a satisfactory set of conditions for the applicability of the word ‘good’. Let us assume for the sake of argument that this set contains the characteristics A, B and not-C. Does ‘good’ then mean ‘A, B and not-C’? If it did, the question ‘This television set has the characteristics A, B and not-C, but is it also good?’ would not make any sense. It would be comparable to ‘Henry is an unmarried man of marriageable age, but is he also a bachelor?’ Asking the latter question merely illustrates that the questioner does not know what ‘bachelor’ means. But does the person who asks ‘This television set has the characteristics A, B and not-C, but is it also good?’ similarly demonstrate that she does not know what ‘good’ means?
Moore famously claimed it did not. It is an ‘open question’ whether something that has the characteristics A, B and not-C is also good. And because it is an open question, ‘good’ does not mean ‘A, B and not-C’. And – this is Moore’s main point – this open question argument can be applied to any substitution for A, B and not-C, with only one exception, namely the characteristic ‘good’ itself. Goodness cannot be defined in terms of ‘natural’ characteristics and any attempt to do so would amount to committing the ‘naturalistic fallacy’.
Originally the term ‘naturalistic fallacy’ was only used to denote the attempt to define evaluative terms in non-evaluative (‘naturalistic’) ones, but in the course of time it has also be used for any attempt to derive normative or evaluative judgments from purely non-normative and non-evaluative premises. An example would be the argument ‘Everybody approves of this article, so it must be good.’ That everybody approves of this article is a matter of fact, and the sentence that expresses it is purely descriptive and non-evaluative. Since being good is an evaluative characteristic, it would then be a naturalistic fallacy to derive from only the sentence that everybody approves of this article that the article is good.
There is a lot to be said about – and against - Moore’s argument but it is a fact that it has been quite influential during the 20th century and that many a derivation of normative or evaluative conclusions has been accused of committing the naturalistic fallacy and therefore being inconclusive. As we will see, the desire to avoid the naturalistic fallacy strongly influences the way in which value judgments are to be justified.
According to a popular view, the universe exists in a way which is independent of human beings, the human mind, and human knowledge. Of course, human beings and human minds are part of the universe and in that sense the universe is not independent of human beings etc. But it does not depend for its existence on our knowledge of it, on the minds that produce this knowledge, or on the humans whose minds are at stake. This view is so widely shared that only philosophers dare to cast doubt on it.
The universe is filled with things which exhibit characteristics and mutual relations. For instance, the universe contains mountains which are high, and human beings, the one taller than the other. That these things exist, have characteristics and stand in relations towards each other can be described by means of declarative sentences such as ‘The mountain is high’ and ‘Joan is taller than Geraldine’. If these sentences describe the universe as it actually is, if the description ‘corresponds to the facts’, the sentences are true; and otherwise they are false. Because the facts are assumed to be mind-independent, the truth and falsity of declarative sentences are also assumed to be mind-independent.
All of this seems so obvious that one easily overlooks some complications. Value judgments constitute one of these complications. The sentence ‘Joan is prettier than Geraldine’ looks very much like ‘Joan is bigger than Geraldine’. At first sight it seems to describe a fact. But is there a fact of the matter concerning the relative beauty of persons? Is there beauty to be found in the universe in the same manner in which length is to be found? Is it possible to measure beauty, more or less in the same way as we can measure length? Many have thought that these questions should be answered negatively. The fact that beauty resembles length in the sense that they are both characteristics which somebody or something can have, is merely a trick that language plays upon us. Grammatically, ‘beauty’ and ‘length’ function in similar ways, but if we look at the underlying reality, they are very different. Length is to be found in the mind-independent universe, beauty is not.
So what do we state if we say that a soccer match or a television set is good, or that Joan is prettier than Geraldine? One set of ethical theorists held that such ‘statements’ are not really statements at all, but merely ways to express our feeling or attitudes about things. These theorists have become known as ‘non-cognitivists’, because they held that evaluative judgments have no cognitive content. These judgments would not give us any information about the universe; in fact they would not aim to inform at all, but merely to recommend, or to express a positive or negative attitude or feeling.
The view that evaluative judgments are little else than expressions of feelings, or recommendations and that a sentence such as ‘This is a good television set’ is ‘really’ a kind of recommendation, rather than the description of a fact has been attacked by Geach and Searle. Basically, their criticism boils down to it that the speech act which can be performed with a sentence does not determine the meaning of the sentence. That a sentence can be used to recommend does not mean that its meaning is ‘recommending’. The sentence ‘This is a good television set’ means the same if it stands on its own as when it is used the conditional sentence ‘If this is a good television set, I should buy it’.
More in general, the meanings of sentences should not be looked for in the speech acts that are typically performed by the use of these sentences. The version of non-cognitivism that attempts to do so, is for this reason misguided.
It is however possible to give non-cognitivism a twist and to interpret it as the view that evaluative judgments do aim to give us information, but not to inform us about the mind-independent universe. Evaluative judgments would be true or false, not depending on a mind-independent reality, but on the outcome of rational argumentation. This view has become well-known under the heading of ‘constructivism’. In legal theory, constructivism can be found in, amongst others, the work of Dworkin and Smits. 
Support for the constructivist view can be found in the phenomenon of supervenience. If a value judgment is to be correct, there must be reasons for it. Take for instance a good soccer match, or a beautiful piece of music. It is impossible that a soccer match is good, or a piece of music beautiful, just for no reason. There must be something that makes the match good, and the music beautiful. It may be hard to indicate and there may be disagreement amongst people about what these good, respectively beautiful making characteristics precisely are, but despite this possible disagreement and lack of clarity, there must be reasons that make a good soccer match good and a beautiful piece of music beautiful. This ‘must’ is a logical must, and it has to do with what philosophers call the ‘supervenience’ of valuation.
A characteristic or phenomenon is supervenient if it depends completely on one or more other characteristics or phenomena (which are then ‘subvening’ characteristics or properties), but if it does not coincide with them. An ordinary object has several characteristics which are independent of each other. For instance, the monitor before me is more or less rectangular, has a black border with white letters on it, and was made by a Korean firm. Each of these characteristics might have been different without any change in the other characteristics. This means that none of the mentioned characteristics supervenes on any of the other, or on a combination thereof. Suppose now that this monitor is suitable for my purposes. It is not possible to take the suitability of the monitor away, while leaving all its other characteristics as they were. Taking away the suitability must take the shape of modifying one or more other characteristics. That makes suitability a supervenient characteristic.
A legal competence is also an example of such a supervenient phenomenon. Legal competences are often attached to, supervene upon, a particular status. The competence to dispose over a good is attached to the ownership of this good (and to a few other phenomena), but does not coincide with the ownership. The competence to create emergency regulations is attached to being a mayor, but does not coincide with it. The competence to pass legal verdicts is attached to being a judge, without coinciding with it. For the presence of all these competences there must be a reason lying in the status of the person possessing the competence in question and if this reason (the status) is lacking, the competence does not exist.
What holds for most competences also holds for goodness, beauty and the presence of any other value: there must be a reason for their existence and this reason is to be found in the characteristics that subvene the value, and upon which the value supervenes.
That value characteristics supervene on other characteristics makes judgments about them ‘universalizable’. Every soccer match that has exactly the same characteristics as a particular good match must be a good one too. The same holds for identical pieces of music: if one of them is beautiful, all the others must be beautiful too. There can be no difference in valuation without an underlying difference in non-evaluative characteristics.
So if a particular soccer match is good because both teams play in an offensive style, all soccer matches where both teams play in an offensive style will be good too. One may argue that things are not that easy. It is well possible that the players in the good match were technically gifted, while the players in a second match were not, and that therefore the second match was not as good as the first, or even bad. That may be a good argument, but this does not show that value judgments are not universalisable, but only that they should be universalized in the correct way: all soccer matches in which both teams play offensively and where all the players are technically gifted are good soccer matches. Obviously it is still possible to think of other characteristics of soccer matches which might make a difference. But the reaction would invariably be to include those characteristics, whether they be positive or negative (no foul play, for instance), in the set of relevant subvening characteristics. And when that has been done, the value judgment can be universalized.
This process of amending the set of relevant subvening characteristics may be difficult and time consuming, and every stage in it is then merely an approximation of the final result in which all aspects that are relevant for a supervenient characteristic are given. The development of case law on a particular legal question, for instance, whether there is liability for defects in a product, illustrates this well. The first case mentions some characteristics which support the supervenient legal judgment about the case. Subsequent cases can serve to broaden the set of cases for which this judgment applies, or to distinguish cases. At every stage in this development one or more sets of characteristics are available which are, according to the then accepted view, sufficient to support the legal judgment in question. The theoretical end point of this development, which may never be reached in practice, is a set of one or more rules without exceptions, each of which specifies by means of its conditions a sufficient condition for a particular legal judgment, e.g. whether the producer of a good is liable for damage caused by a shortcoming of this good. This sufficient condition is the subvening characteristic that underlies the legal liability.
That value judgments are universalizable means that for every value judgment it should be possible to identify a set of characteristics, positive and/or negative, on which the valuation supervenes, and which is such that all cases in which this set of characteristics is present, the same value judgment is correct. For instance, if a particular soccer match is good because both teams play offensively, the players are technically gifted and there is hardly any foul play, then all soccer matches with these characteristics are good too. It may be hard to formulate the set of conditions which provides sufficient support for a value judgment, but the supervenient nature of valuation makes that for every value judgment there must be such a set.
The result of universalizing a value judgment is a kind of ‘rule’, called a ‘standard’. One such standard might be that soccer matches in which both teams play offensively, the players are technically gifted and in which there is hardly any foul play are good ones. Another, fictitious, example would be that a contract party acts in good faith if, in deciding how to perform the contract, he assigns the same weight to the interests of his contract party as to his own.
Notice that the standards formulated by universalizing a value judgment provide sufficient conditions for a particular valuation, but not always a necessary condition. There may be more than one way in which a soccer match is good, and more than one way to perform a contract in good faith. For each of these ways there will be a standard (‘rule’), and the conditions in these standards are individually necessary and unitedly sufficient for the valuation on the basis of this standard.
After the digression on the nature of valuation, it is time to return to syllogistic justification. For a while we will focus on the justification of value judgments only. A syllogistic justification of value judgments is an argument with as conditional sentence a standard and as facts the characteristics of what is to be evaluated. If these characteristics satisfy the standard, the evaluative judgment logically follows and is justified. This simple seeming picture of syllogistic evaluation hides some complications, however.
In a good argument, the conclusion ‘follows’ from the premises. But what does that mean? In logic there is one standard for when a conclusion follows from the premises that is above all discussion, and that is the standard of deductive validity. Deductive validity of an argument is defined in terms of the truth of the premises and the conclusion of the argument. An argument is deductively valid if and only if it is (logically) impossible that all the premises of the argument are true, while the argument’s conclusion is false. This means that this standard can only be applied to arguments in which all the premises and the conclusion have a ‘truth value’, meaning that they are either true or false. If the conditional premise is not a statement with a truth value, but a standard, the argument cannot be deductively valid. Not because it is deductively invalid, but because the very notion of deductive validity is not defined for arguments that have premises without a truth value.
Even though the standard of deductive validity is usually not applicable, we seem to be capable to evaluate arguments as good ones, if the conclusion somehow follows from the premises, and bad ones, if the conclusion does not follow. Apparently we apply different standards for validity next to the one of deductive validity. What these standards are, seems to vary with the field of the argument. Standards for medical diagnosis differ from the standards that engineers use to calculate how strong the foundations of a building should be, which in turn differ from the standards applied by lawyers to determine whether the defendant is guilty, or whether the debtor acted in good faith.
In all these fields it is possible to distinguish between good and bad evaluative arguments, but the standards used to distinguish good arguments from bad ones differ. Only by abstracting from all the differences it is possible to formulate a universal standard: an argument is good if, in the light of the premises, it is rational to accept the conclusion of the argument, rather than to reject the conclusion or to suspend judgment. That leaves the question open what counts as rational, which again would depend on standards that are field-dependent.
This leads us back to the universalizability of value judgments. If a value judgment is universalized, the result is a standard that specifies what are the good or bad making characteristics of, say, a soccer match, an audio set, or a legal rule. This standard is needed to support the value judgment and can at the same time be used to distinguish between good and bad evaluative arguments. Universalization of value judgments is a kind of inductive reasoning, meant to establish the standard underlying the value judgment. Evaluative reasoning is to apply an evaluative standard in order to support a value judgment. Although such an argument cannot be deductively valid, because a standard is not true or false, the style of reasoning involved in the application of an evaluative standard has much in common with a traditional syllogism.
Schematically the relation between universalization and the application of a standard can be represented as follows:
An example of universalization was given in section 4.4. Let us suppose that soccer match A was evaluated as good because both teams played offensively, the players were technically gifted and there was hardly any foul play. From this evaluation, based on these reasons, we can by means of universalization ‘derive’ the standard that a particular soccer match is good (sufficient condition) if both teams play offensively, the players are technically gifted and there is hardly any foul play. This same standard can then be applied to match B which has the same relevant factual characteristics and then the argument in which this standard is applied counts as justification for the value judgment that match B is good too.
Notice that this line of argumentation has many similarities to an argument by analogy, but that it is not such an argument. It is not argued that soccer match B is similar to match A and that therefore match B must be good since match A was good. That would be reasoning by analogy, based on the comparison of soccer matches. The line of argumentation presented here is more explicit and makes the standard which is applied explicit.
Although the formulation of a standard that supports value judgements can be found by universalizing an existing value judgement, this procedure does not guarantee that the thus found standard is ‘correct’. The ‘correctness’ of the standard depends both on whether the value judgment that formed the starting point of the universalization was correct itself, and on whether the proper good and bad making characteristics were identified by means of the universalization. We will soon return to the correctness of the standard, but first it must be pointed out what precisely the role of the standard’s correctness is.
To see this role in perspective, it is useful to return to the original model of justification in which a descriptive judgement was justified on the basis of a descriptive major premise and a true description of the facts of a case or an object. As example of this kind of justification we used the argument
If two witnesses testified that Jerry murdered his mother in law, then Jerry did commit that murder.
Both Jerry’s wife and his father in law testified that they saw Jerry murder his mother in law.
Jerry murdered his mother in law.
The point of justifying the conclusion is to show that the conclusion is correct, and correctness in connection with descriptive sentences is truth. This conclusion can therefore be justified by showing that it is true. And that can in turn be shown by means of a deductively valid argument that has two true premises. The minor premise is a description of the case facts and for the sake of argument we can assume that this description is true and that Jerry’s wife and his father in law actually testified that they saw Jerry murder his mother in law.
If the argument is to be a successful justificatory argument, its major premise must be true too. This is somewhat harder to establish, because the major premise is a conditional sentence and the truth of conditional sentences cannot be established by means of sensory perception alone. We will return to this complication later (section 6.4) but apart from that we can establish that the major premise is a sentence with a truth value. If the major premise is true, the argument, which is deductively valid, proves the truth of the conclusion and in that strong sense justifies the conclusion.
If we move from the justification of a descriptive conclusion to the justification of a value judgement, two things change. First the major premise becomes a standard, and standards are not true or false. And second, it may at least be doubted whether a value judgment can be true or false. If a value judgment has a truth value, it is in the eyes of most people not because it corresponds to an objective, mind-independent fact. These two differences make that the simple picture of justification as a deductively valid argument with two true premises cannot apply to the justification of value judgments.
It is possible, however, to save much of this simple picture by broadening the notion of justification. A conclusion counts as justified if it follows from a major and a minor premise which are both justified themselves. There may be several ways in which a premise can be justified, but one of them is being true: a premise counts as justified if it is true. If this step is taken, the notion “being justified” takes the place that was held by ‘being true’ in the more simple picture. And yet the role of truth is maintained by counting any premise that is true as being justified. This adapted version of the deductive picture of justification underlies explicitly or implicitly many discussions about the justification of legal judgments. 
A standard cannot be justified in the sense that it is true, since standards have no truth values. On the assumption, which will be challenged later (section 7), that a premise of a justificatory argument can only be justified by being true or by being the conclusion of another justificatory argument, it follows that if a standard is to be justified, it must be as the conclusion of a justificatory argument. This opens the question which kinds of arguments can justify a standard.
A standard is not correct in the sense of being true, but rather in the sense of being a good standard. That a standard is good is a value judgement itself, and can only be justified by means of an argument that applies a standard. In general, an argument that justifies a standard must itself have a different standard as its major premise. For instance, the standard that good laws must be understandable can be justified by the argument that good laws must be capable of guiding conduct and that only rules which are understandable can guide conduct.
The insight that an argument that justifies a standard must itself have standard as its major premise immediately leads to a complication. The standard which is used to justify another standard must itself be justified too. And this justification apparently must consist in another justificatory argument with a standard as its premise, and this standard must itself be justified by means of still another argument which …; and so on. It seems that the justification of a value judgement leads to an infinite chain of standards that justify standards.
Can such an infinite chain be avoided? That is possible, but only at some costs. In the following subsections a number of possibilities will be discussed. None of them is fully satisfactory, and therefore section 7 discusses an alternative for the foundationalist approach to justification.
One solution for the problem of a threatening endless regress is not to use ‘good’ standards but existing standards. In fact, this is a solution which will appeal to many lawyers, because it is used in law to determine what people should do. Instead of requiring from people that they do what is morally required or what is the best possible act, it ‘only’ requires that people comply with the rules that actually exist as part of the law. This not only avoids the endless regress that may be involved in the justification of value judgments, it also prevents continuing discussions about what is morally right or what is the best thing to do. Avoiding such discussions is valuable in itself.
Analogous to the application of positive law as standard for behaviour, one can appeal to positive standards as the basis for value judgments. The DIN standard that was mentioned in an earlier example illustrates this possibility. It has actually been adopted as standard that must be satisfied by hifi-sets in order to qualify as good ones. Similar standards exist in sports such as gymnastics where they are used to evaluate the quality of for instance free exercises. These standards exist in the sense that they are actually accepted within a social practice: they are being used, they can be cited as reasons for value judgments, and if they are not used, this is a reason for criticism. The existence of such standards can be established as a matter of social fact, and the empirical social sciences are in principle able to give ‘hard’ outcomes as to the existence of a particular standard. The judgement that a particular standard exists is true or false, which means that the justificatory chain can end with a standard that is considered to be ‘justified’ in the sense that it actually exists.
Related to the use of standards that actually exist is the use of standards that are ‘presupposed’ by an existing social practice. Law provides a good example again. There are legal rules which exist as a matter of fact, for instance because they were created by means of legislation or were adopted by the judiciary as case law. These rules apply only to the cases that satisfy the rule conditions, the cases to which the rules are applicable. However, it is possible that some legal principle underlies a set of these rules. Examples would be the principle that third parties in good faith deserve protection, or the principle that nobody should profit from his own wrongs. These principles do not exist in the sense that they are normally used when the opportunity presents itself. In ordinary cases the rules in which the principles are incorporated are used, not the principles themselves. However, the use of rules based on principles illustrates that in some sense these principles do ‘exist’. And if a case arises to which no rule applies but which nevertheless falls under such a presupposed principle, it may be possible to justify the judgment about this case by invoking the principle. This holds for legal judgments, but also for evaluative social practices. The existence of ‘presupposed standards’ is something that can in principle also be established by means of social science, be it that it not only requires the establishment of which rules belong to a social practice, but also the additional interpretational step of finding the principles (rationales) behind the rules.
Somewhat related to standards that are actually being used or which are presupposed by an existing social practice are policies that have been adopted by a person who, or a body that, is competent to set standards. For instance, a legislative body may formulate a policy to determine under which circumstances new legislation is desirable. The existence of such policy standards can also be established as a matter of social fact.
The invocation of existing rules and standards, explicit or presupposed, or of policies, has the clear advantages of increased objectivity and cutting off potentially endless lines of regress of standards underlying standards. There are also disadvantages, however. Actually existing standards may be bad ones, as is sadly illustrated by the actual use of standards according to which coloured people are ‘worse’ than non-coloured ones, or women inferior to men. Is a value judgment based on such a bad standard really justified?
The same question may also be approached from the perspective of the naturalistic fallacy. Is the argument from ‘This standard is actually used’ to ‘This standard justifies the value judgments based upon it’ not a clear instance of the naturalistic fallacy? We encounter this discussion in the law too, when the question is raised whether bad ‘laws’ are really law, or whether they ‘really’ deserve compliance.
This complication can be solved by means of a definition. It is possible to define a social practice in terms of the standards that are actually being used within this practice. For instance, one may define the moral point of view in terms of the standards that are in a particular society actually used for moral evaluation (positive morality), or the legal point of view in terms of the legal sources that are recognized in a particular legal system (positive law). Then it can be established ‘objectively’ what is morally right or wrong, and what legally is the case. This definitional approach has the disadvantage, though, that the question may pop up whether one should ‘really’ evaluate acts and situations morally, or whether one should ‘really’ live in accordance with the law. The more objective standards become, the more easily the question will arise whether these standards should ‘really’ be used.
The use of actually existing standards, whatever shape they may take, gives rise to the question why these positive standards should be used. If that happens we are back at the phenomenon that the use of a standard only justifies a value judgment if the standard is itself justified by means of a standard applying argument. How to avoid the infinite regress of standards underlying standards to which this phenomenon threatens to lead?
An alternative for the invocation of ‘positive’ rules and standards seems to be an appeal to self-evidence. Such an appeal is not often made explicitly, but some arguments seem to presuppose it. An example of such an implicit appeal to self-evidence as a means to stop a chain of infinite regress is the ‘defence’ by John Stuart Mill of the Greatest Happiness Principle, the principle that those acts are best that maximize the total amount of happiness. Mill starts in the introductory chapter of his booklet Utilitarianism to claim that ultimate ends, such as happiness, cannot be proven in the ordinary sense, because such an ordinary proof would require the availability of an even more ultimate end which by definition is lacking in the case of ultimate ends. Mill gives, amongst others, the example of the medical art which is proved to be good with an appeal to health, but that it is not obvious how to prove that health is good. In the fourth chapter of utilitarianism, Mill ‘proves’ the utilitarian doctrine that happiness is desirable from the fact that each person actually desires to be happy. Being aware that this does not sound very convincing, Mill adds that this is ‘… not only all the proof which the case admits of, but all which it is possible to require, …’.
The problem with such appeals to self-evidence, as Mill apparently makes in this very brief argument why happiness is desirable, is that it only convinces those who were already convinced. Those who were not convinced already will not consider such an argument a real justification.
In general, the attempt to find a firm foundation for standards fails because the search ends in an infinite regress, or in some way of dogmatizing one or more particular standards. At first sight this problem seems peculiar to the justification of value judgments, because the justification of a value judgment needs a ‘good’ standard as a premise. On closer inspection, the problem is not confined to value judgments, however. We have already encountered an ‘ordinary’ justificatory argument with two true premises, the first of which was a conditional sentence. It was then remarked that it is not so easy to show the truth of a conditional sentence. We lack the senses to perceive conditional facts. The same counts for universal statements such as the statement that metal bars expand when heated. We assume that such law-expressing statements are true or false, but it is far from obvious how this truth or falsity can be established.
This problem of establishing truth does not only play a role in connection with conditional and law-expressing sentences, but also in the case of ‘ordinary’ descriptive sentences, such as ‘This is a metal bar’. It is one question whether this sentence is true, and another how it can be shown to be true. In a number of cases this may seem simple: the truth of a sentence is ‘given’ to our senses. We can see a horse before our eyes, and we can hear the music play. But even in these simple cases there are grounds for doubt. Does the fact that we see a horse prove that there is a horse? In a sense, it does, because if there were no horse we could not have seen a horse. However, this observation immediately backfires: did we really see a horse, or did we only believe to see one? Does the experience of seeing a horse provide one with sufficient evidence to be justified in believing what is apparently experienced? And the justificatory issue is back: even on the truth of the sentences that are obviously true because their truth is ‘given’ to our senses doubts may be cast. Is our sensory evidence really sufficient evidence? In case somebody is tempted to point out that if anything is obvious, this is, the equally ‘obvious’ reply is that this is a typical example of an appeal to self-evidence, an appeal that was found to be unattractive in case of other standards. Why would such an appeal suddenly be attractive if the standard is that beliefs are justified if they are based on sensory perception? This point can be made in a more general way: there is no information that is ‘given’ to us, either through our senses, or to some other faculty that may provide us with knowledge. The idea that there would be such information has become known as the ‘Myth of the Given’.
If the hope is abandoned that true descriptive sentences are in a privileged position with regard to justification, because sensory perception provides us with the necessary justification for our beliefs in them, it turns out that it does not matter whether a premise is true. What counts is whether the acceptance of the premise was justified. This holds for standards just as well as for descriptive premises. Being true cannot take the place of being justified, as was suggested earlier (section 5.2). All premises of a justificatory argument must themselves be justified, and this recursion – the definition of ‘being justified’ in terms of being justified - cannot bottom out on premises that are justified because they are true. The problem with a foundationalist approach to justification is even worse than it seemed at first. It does not only relate to standards which lack a truth value; it relates to all premises of justificatory arguments. Apparently these premises either must be assumed dogmatically, or the justification of any judgment, including any value judgment, turns into an infinite regress.
The enterprise to justify value judgements seems to be hopeless, but happily it is possible to challenge the syllogistic model of justification from which the foundationalist problems result. The traditional alternative for the foundationalist approach to justification is to look at justification in a coherentist fashion.
The difference between the syllogistic/deductive model of justification and the coherentist one can well be represented by means of a metaphor. According to the syllogistic/deductive model, justification is like a pyramid, with at the top the position that is being justified, in the present context a value judgment. At the bottom of the pyramid are the ultimate premises of the justificatory argument, and in between are the intermediate conclusions that function as premises for the arguments built on top of them. The problem of the syllogistic/deductive model is that the foundation of the pyramid seems to float in the air.
The most appropriate metaphor for the coherentist approach to justification is that of a bundle of standing sticks that remains in balance because the sticks support each other. No stick remains unsupported, but in a sense the chain of support is circular because the one stick is supported by another stick which is supported by still another stick …. etc. until the first stick reappears again as support for a stick that through a series of other sticks indirectly supports the first stick.
The intuition behind a coherentist view of justification is that something cannot be justified in isolation, but only as an element of a larger set. This set is to be coherent, and all the elements of the coherent set are justified for the reason that they belong to a coherent set. There are some obvious objections against this view, and they will be mentioned and briefly discussed in section 7.2. However, first the idea of justification through coherence needs to be elaborated.
Before continuing the discussion, it is useful to introduce some technical terms. The elements that belong to a coherent set can be very diverse. They include not only descriptive sentences that are true or false, but also value judgments of which some people doubt whether they are true or false, and standards that support value judgments and which are certainly not true or false. They may also include rules, both mandatory rules as they abound in the law, but also rules of inference such as the ones used in logic, which allow one to draw conclusions from sets of premises. Rights and interests can also be elements of a coherent set, just as values and policies. Some of these entities, in particular descriptive sentences, and maybe also value judgments, can be believed. Others, such as rules may exist in the sense that they are valid. Again others, such as policies, can be adopted or accepted. It is useful to have one term that can be used for all elements of a coherent – or not yet coherent – set, and I propose to use the term ‘position’ for this purpose. Positions can then be said to be ‘accepted’ by the person who believes them (in case of descriptive sentences), considers them to be valid (in case of rules), or to exist (values, rights). So a set that aspires to be coherent consists of positions that are accepted by the agent (person, or organization) holding this ‘position set’. The opposite of being accepted is being rejected. A rejected element is not an element (anymore) of the position set held by an agent.
The metaphor of sticks
supporting each other that underlies coherentism may have its attractions,
but it is far from easy to substantiate the idea of coherence. What makes a
position set coherent? A first demand is that the elements of the set do not
‘contradict’ each other; a coherent position set must be ‘consistent’. The
words ‘contradict’ and ‘consistent’ are within quotes, because their meanings
must be stretched somewhat if they are not only to be applied to descriptive
sentences, which are true or false, but also to other positions such as values,
rules. This is not the place to elaborate how different kinds of positions can be (in)consistent. Suffice it to claim that a position set is not consistent (inconsistent) in the sense used here, if and only if at least one of its elements should rationally be rejected in the light of (the acceptance of) its other elements.
An inconsistent position set is not coherent, but does consistency suffice to make a set coherent? Many have thought that mere consistency is too weak a demand to make for coherence. It should not only be required that the elements of a coherent set do not contradict each other; they should also support each other. However, it turned out to be quite difficult to define what this ‘support’ might amount to. Again, this is not the place to elaborate that theme, and I will confine myself to proposing a possible solution to the problem of defining support. This solution is to drop the demand that the elements of a coherent position set should support each other and replace it by the even more challenging demand that a coherent position set be ‘comprehensive’.
To see what this demand implies, and why it is so demanding, we must return to consistency again. The requirement that a set be consistent becomes more demanding, the larger the set is. A single elementary sentence, such as ‘This is a good rule’ cannot even be inconsistent. To make the set inconsistent, it is at least necessary to introduce a second sentence; ‘This is not a good rule.’ The more sentences are added to a set, the more opportunities arise for the occurrence of contradictions. Where a small set can hardly be inconsistent, a very large set easily becomes inconsistent. Therefore, the requirement that a set be consistent is more demanding if the set is very big than if the set is only small.
Here is where the requirement of comprehensiveness enters the picture. A comprehensive set of beliefs consists of everything that an agent believes. A comprehensive position set is even bigger, since it does not only contain beliefs, but also all the values, rules etc. which the agent accepts. The requirement that such a set be consistent is very demanding, because it amounts to the demand that the agent is completely rational and does not accept anything that rationally should not be accepted in the light of everything else that the agent (rationally) accepts. In fact, this requirement is so demanding that no human agent will ever be able to fully meet it.
If a comprehensive position set is consistent, this means that no decisive criticism of any element in it is possible on the basis of the other elements of the set. An agent holding such a set would seem to be fully justified in doing so, and what more could be asked? ‘Truth’, some may be tempted to answer, but as was argued above (section 6.4), what counts for justification is not whether a belief is true, but whether it is justified. Therefore it seems that an agent who holds a consistent comprehensive position set is justified in doing so, and this justification not only applies to the set as a whole, but to every position in it. So an agent would be justified in accepting a value judgment if this value judgment is an element in the comprehensive position set held by this person, and if this position set is consistent.
This is not the whole story, however. It may be the case that an agent’s position set is consistent, but that it lacks some elements which rationally should be included. Suppose, for instance, that an agent both accepts the standard that soccer matches in which both teams play offensively, the players are technically gifted and in which there is hardly any foul play are good ones, and that in the match between Liverpool United and Schalke 04 both teams played offensively, the players were technically gifted and there was hardly any foul play. Then this agent rationally should also accept the value judgment that this match between Liverpool United and Schalke 04 was a good one. If this judgment is not yet part of the agent’s positions set, the agent is not fully rational, and the set is not coherent. A coherent set should be consistent, but it should also include everything that rationally should be included given (the rest of) the set. So if, for instance, a particular value judgment follows from a position set, this value judgment should also be in the set. This will be called the requirement that the position set is ‘complete’. Completeness is a requirement on coherent position sets.
Sometimes an agent deliberately refuses to include a position is her position set, even it should rationally be included given the rest of what she accepts. Then the set can also be made coherent by dropping an element. In the above example, the agent may drop one of the beliefs that both teams played offensively, that the players were technically gifted or that there was hardly any foul play. The agent may also stop accepting the standard that soccer matches in which both teams play offensively, the players are technically gifted and in which there is hardly any foul play are good ones. So coherence can not only be restored by adding elements that are lacking, but also by giving up elements which require that some undesirable element should be added to the set.
The attentive reader may have noticed that in the exposition of justification by means of coherence, a change of formulation occurred. The phrase was not anymore whether a belief or standard, or – in general – a position was justified, but whether an agent was justified in accepting this belief, standard or position. This change in formulation reflects a fundamental change in the question that is being answered. When we talk about a belief or standard being justified, it seems that we presuppose that ‘being justified’ is a characteristic that beliefs and standards can possess. We encountered this presupposition in section 5.2 in discussing the replacement of truth by justification for positions like standards which are not true or false. In section 6.4 it was pointed out that even when a position can be true or false, in particular when it is a belief, it is not the truth that counts, but being justified. However, it remained unclear what kind of characteristic ‘being justified’ is. Truth can be defined in terms of correspondence to the facts, but how should we define ‘being justified’? Is it a way of being right, just like truth, but then more abstract to make it potentially applicable to entities that have no truth values, such as standards or rights? If so, it remains unclear in what way entities without a truth value can be right. What does a justified standard have that a non-justified standard lacks?
If it is not the position that is, or is not, justified, but rather the agent in accepting a particular position or position set, it becomes easier to see what justification amounts to. An agent can be said to be justified in accepting a position if accepting this position is the rational thing to do. That would, on a coherence theory of justification, be the case if this position is an element of a coherent position set held by this agent. The agent is then also justified in holding this position set, for the reason that the set is coherent (comprehensive, consistent and complete).
The view that it is the agent who is justified, and not the positions accepted by the agent, fits a coherentist approach to justification, since coherentism tends to make justification agent-relative. This has to do with the requirement that a coherent position set be comprehensive, since a comprehensive position set consists of every position accepted by an agent. This definition makes comprehensiveness and therewith coherence agent-relative. This relativity is in the eyes of many a disadvantage of coherentism. If two agents disagree about a value judgment, what does it help if the one agent is justified in holding this value judgment, if at the same time the other agent can be justified in rejecting it?
Often the focus of the criticism is not so much on the agent-relativity, but rather on the isolation of the positions in a coherent theory from the facts in the world. A positive metaphor for coherentism is that of a bundle of sticks which can all stand because of the support they receive from the other sticks. However, Haack borrows from C.I. Lewis the more negative metaphor of a bunch of drunken sailors who can only remain upright because they hang together. The problem is that according to coherentism, the justification of any position, inclusive beliefs about the external world, only depends on other positions held by the agent. This not only makes coherence and justification agent-relative, but there also seems to be no guarantee that the positions in the set are in any way related to the facts. Sosa describes in this connection a thought experiment. Suppose than an agent holds a particular coherent position set, created under the influence of the facts in the world. All beliefs in the set are true. Through a miracle all the facts in the world change, and all beliefs in the position set have become false. But allegedly, the position set itself would still be coherent. Can this coherence still justify the beliefs and other positions in the set?
Apparently the correct answer to this question is negative. However, the question should not be whether the positions are justified. We still lack an adequate theory of what being justified means in relation to positions. The proper question is whether the agent is justified in accepting the positions in her position set. If the agent in the thought experiment is not aware of any change in the facts – which is highly unlikely, because some of the facts concern his own mental states – the agent would still be justified in holding the belief set. Why should she modify any positions if the set satisfied the highest demands of rationality and nothing came up to create the impression that this is not the case anymore? Much more likely is the situation that the agent is aware of at least some of the changes in the world. That would mean that some positions in her set have changed spontaneously, that is because of factors outside the position set itself. Most likely the set is not coherent anymore and the agent would not be fully rational in returning to the set as it originally was.
The conclusion from this all is that if we take the agent as the entity that is to be justified, justified in accepting positions, the fact – if it is one – that a position set can be completely disconnected from the world is no decisive objection against coherentism. It may make coherentism less attractive for some, because the claim that an agent is justified in subscribing to a particular value judgment may be considered as less ambitious than the claim that the value judgment itself is justified. Those who want justified value judgments may not be satisfied with agents being justified in accepting these value judgments. However, since it is not clear at all what a justified value judgment would be, it is questionable whether those who strive for justified value judgments make any sense.
That does not end the discussion. What if two agents are both justified in accepting conflicting value judgments? Would they both be ‘right’ then? It depends. For some value judgments this might be the appropriate solution, for instance for value judgments like ‘Apple-sauce tastes better than broccoli’, if both agents agree that this is merely a matter of taste. But then the value judgment ‘Apple-sauce tastes better than broccoli’ is being considered as shorthand for ‘I prefer apple-sauce above broccoli’. If two agents disagree about the truth of this sentence as applied to themselves, they do not disagree after all.
If the agents do not believe that a particular value judgment, for example that one rule is better than another, is merely a matter of taste, they will probably assume that they have good reasons for sticking to their own value judgments. Then, most likely, the rational thing to do is to listen to the reasons of the other agent. This may change one’s position set, and possibly also one’s value judgment. And if not, there may be reason to continue the search for what is the best value judgment. Although there is no guarantee that this process makes that any of the agents rationally should (rather than ‘will’) change his mind, most likely it will make it rational for an agent, or for both, to change their minds.
A coherent position set is a complete, consistent and comprehensive – that also means: actually held – position set. In reality such things do not exist. The idea of a coherent position set is a theoretical construct, useful for defining what it means for an agent to be justified in accepting a particular position, such as a value judgment. If coherentism is to be used as a real test on the justification of positions, a less demanding version should be devised.
One such less demanding version is defeasible coherentism. The starting point is again the idea that justification should be based on coherence and that a coherent set is to be consistent and complete. However, the requirement that this set is comprehensive must be dropped, because a comprehensive set would be unmanageable. Therefore the position the acceptance of which is to be justified does not have to be an element of the full comprehensive set, but merely of a ‘representative’ subset thereof. This subset should be consistent.
The crucial extra demand on the subset is that it is ‘representative’. A representative subset is a subset which does not give a wrong picture of the full set where the position that is to be justified is concerned. Ideally, the subset should only contain this position if the full set would also contain it. Obviously this cannot be established without having disposal over the full set, and that is precisely what will not be available. The solution to this problem can be found in the use of meta-knowledge. Somebody can hold a belief which, as far as she can see, fits in the set of all her positions. This means that she believes that her belief fits in her actual comprehensive position set. She does not know whether this meta-belief about the adequacy of her subset is true, but she nevertheless has it. Moreover, as long as she does not have counter-evidence, she is defeasibly justified in sticking to it. Not having counter-evidence means in this connection: not being aware of counter-evidence and not having counter-evidence in the same defeasible sense that the subset of her positions of which she is aware contains reasons not to trust it.
An example may make this clearer. A public prosecutor believes that John is a thief and that the rule exists that thieves are punishable. Given this limited position set, she should adopt the belief that John is punishable. Actually she also believes that minors are not punishable, but she has no reason to assume that John is a minor. Therefore she actually believes that her small position set is representative for her actual comprehensive set where the belief that John is punishable is concerned. Moreover, as long as she is not aware of counter-evidence (there is no counter-evidence in her subset) she is justified in sticking to this meta-belief. The subset is defeasibly coherent and therefore this public prosecutor is defeasibly justified in believing that John is punishable. That would become different if somebody informs the prosecutor that John is a minor. Maybe the prosecutor does not immediately believe that, but now she has reason to conduct additional research to rebut what was told her, or – if the information turns out to be correct – to modify her limited position set to make it include the beliefs that John is a minor and that he is not punishable. The belief that John is not punishable should at the same time be removed from the position set then.
This version of defeasible coherentism is an example of a procedural approach to justification, since it makes use of the set of positions that are actually being held by an agent and its transformations in time, and not of an ideal set such as a coherent one. In section 10 we will encounter other procedural approaches to justification.
Often value judgments are the outcome of balancing reasons. Take the value judgment that contract law is better regulated on the level of the EU than on the national levels of the EU Member States. One reason why this value judgment might be correct is that there are many contracts between businesses in different EU countries. There is also a reason against this value judgement, however, and that is that regulation on the EU level is an infringement on the national autonomy of the EU Member States. To determine the correctness of the value judgement, these reasons need to be balanced. It is not so easy to account for the balancing of reasons in a syllogistic model of justification, however. Therefore, the account of syllogistic justification as given above will here be supplemented by a brief description of how justification on the balance of reasons works.
The word ‘reason’ is not always used in the same way by everyone, and neither does it have a single meaning. The following is therefore to some extent stipulative.
Reasons are facts, which can be described by true descriptive sentences. For instance, the fact that it is raining is a reason to carry an umbrella, the fact that the metal bar is heated is the reason why it will expand, the fact that the polygon has four straight corners is the reason why it is a rectangle and the fact that John was treated with disdain is the reason why he started to cry.
As these examples illustrate, there are different kinds of reasons. Some guide behaviour, such as the fact that it is raining. Others cause events, such as the fact that the metal bar was heated. Again others make that something is the case, such as the fact that a polygon has four straight corners. And still others explain events that took place, such as the fact that John was treated with disdain.
All reasons have in common that they are facts that are somehow relevant for other facts. Since relevancy if not a one-time affair, reasons can be universalized. If fact F is a reason for fact P, then normally facts like F will be reasons for facts like P. We have already encountered this phenomenon in section 4.3 when supervenience was discussed. A fact P is supervenient on another fact F if F is (part of) the reason why P obtains. Being good is an example of a supervenient fact, just like the presence of any other value characteristic. This means that if a particular soccer match is good for the reason that it has characteristic C, then if some other soccer match also has characteristic C that is a reason why this other match is also good.
A kind of fact F that is normally a reason for another kind of fact P, does not have to be a reason under all circumstances. There may be exceptional circumstances. For instance, the fact that I promised to give you €50 is normally a reason why I should give you €50. But if I was forced to make this promise, that is a reason why my promise does not constitute a reason in this particular case. Reasons which make that facts which are normally reasons are exceptionally not reasons, are called ‘exclusionary reasons’.
Contributory reasons are reasons that plead into a particular direction, but which may have to be balanced against other reasons pleading in the opposite direction. For instance, the fact that it is raining is a contributory reason to carry an umbrella, which may have to be balanced against the reason that there are more important things to carry.
Decisive reasons decide the issue for which they are a reason, without any need for balancing. For instance the fact that there are three male elephants and four female ones are together a decisive reason why there are seven elephants. There is no room for balancing here. It is well possible to treat the combination of a standard and the facts that match the standard together as a decisive reason for the value judgment based on this standard. For instance, the standard that soccer matches in which both teams play offensively, the players are technically gifted and in which there is hardly any foul play are good ones, in combination with the facts that in the match between Liverpool United and Schalke 04 both teams played offensively, the players were technically gifted and there was hardly any foul play together might be seen as a decisive reason for adopting the value judgment that this match between Liverpool United and Schalke 04 was a good soccer match.
Notice that if a contributory reason decides an issue after balancing, this does not make it into a decisive reason. It remains a contributory reason, be it a contributory reason that has decided the issue.
Sometimes an event occurs which makes that something that was a reason is not a reason anymore. If John promised Jane to give her €50 this is a reason why John should give Jane €50. However, as soon as Jane released John from his promise, the reason does not exist anymore. It is still the case that John made the promise, but this fact has lost its relevance for what John should do. Such events which make that facts that were reasons lose their status as reasons, are called ‘cancelling conditions’. Cancelling conditions should be distinguished from exclusionary reasons, because exclusionary reasons makes that some other facts are not reasons at all, while cancelling conditions make that some facts that were reasons lose their force as reasons, are not reasons anymore.
Any judgment, including a value judgment, counts as justified if there exists a decisive reason for it.
In the case of value judgments there will often be no decisive reasons, and in that case the justification of the judgment must be based on the balancing of contributory reasons.
In this connection a misunderstanding lurks. The term ‘balancing’ may be used for the mental process of making up a decision, but that kind of balancing does not lead to the justification of a value judgment. At best it can explain why a person adopted a particular value judgment, and explanation is not justification.
For the present purposes, ‘balancing’ stands for a logical step in which a conclusion is derived on the basis of three premises. One premise states the contributory reasons that plead for the conclusion. Another premise states the contributory reasons that plead against the conclusion. And a third premise, the ‘weight-premise’, states which of these two sets outweighs the other set. For instance:
The fact that there are many contracts between businesses in different EU countries is the only contributory why contract law should be created on the level of the EU.
There are two contributory reasons why contract law should NOT be created on the level of the EU:
1. that regulation on the EU level is an infringement on the national autonomy of the EU Member States;
2. that national states know better than the EU which contract law is good for them.
The set consisting of the single reason why contract law should be created on the level of the EU outweighs the set consisting of the two reasons why contract law should NOT be created on the level of the EU.
Therefore: It would be good if contract law were created on the level of the EU
It is important to notice that the weight-premise, indicating the relative weight of the two sets of reasons, is a premise of the balancing step, not the result of it. Given this premise the logical step to the conclusion is ‘logical’, as it should be in logic. Often the most difficult step in the reasoning process that leads to a decision is to establish which set of reasons is the stronger one. The weight-premise of the argument only gives the result of this decision making process.
Notice, by the way, that it is not automatically the case that the set with the larger number of reasons outweighs the other set. In the present example, a set containing only a single reason is taken to outweigh a set containing two reasons.
The balancing of contributory reasons takes place between two sets of reasons, the reasons pleading for the conclusion, and the reasons pleading against. This is also the case if a set contains only one reason, or even no reasons at all (the ‘empty set’). The reasons in these two sets consist of all the facts that plead for, respectively against the conclusion. They must be facts, so the sentences expressing them must be true. They must also be relevant, so the relevancy of the facts should not be excluded by an exclusionary reason, or taken away by a cancelling condition.
The determination of which set of reasons outweighs which other set is often a matter of decision making for which no additional reasons can be adduced. In that case the weight premise is a premise of the full argument indeed, and not an intermediate conclusion. However it is sometimes possible to argue why one set of reasons outweighs the other, and then the weight premise is an intermediate conclusion in the full argument.
The simplest case is when there are only two reasons, the one for a conclusion, the other against it, where these two reasons somehow fit on a single scale. The attractiveness of the law and economics approach is that it provides a scale, that of efficiency, on which many legal reasons fit. The same counts for utilitarianism, which measures everything on the single scale of the maximization of happiness. It may be a reason pleading for the creation of contract law on the level of the EU that there are many contacts between agents in different EU countries. Uniform contract law then leads to less transaction costs and that leads in turn to larger efficiency; at least that is arguable. That would, on the assumption that efficiency is desirable, be a contributory reason for regulation of contract law on the EU level. However, uniform contract law may force many companies to adapt their standard contracts to the new EU law. That is costly and less efficient and therefore a reason against regulation on the EU level. In this example, efficiency constitutes a common scale on which both reasons can be positioned. Moreover, it can, at least in theory, be computed whether the gain in efficiency of the decreasing transaction costs is bigger than the loss in efficiency caused by the necessary modifications of the standard contracts. The outcome of this computation then decides which reason outweighs the other.
Very often, however, such a common scale is lacking, and then the question arises how at first sight very different reasons can be balanced against each other. In this connection the term ‘incommensurability’ has become popular: reasons that do not fall on a single scale would be incommensurable. In the present author’s opinion, incommensurability is not very problematic. Some facts exist without there being any reason for their existence. Why would not the fact that one set of reasons outweighs some other set belong to this category of facts that can exist without a reason?
Interestingly, it is sometimes possible to argue about the relative weight of sets of reasons even if initially there were no reasons why one set of reasons would outweigh another. Precedent plays a role in this connection. If one set was once taken to outweigh another set, this may count as a reason why the ‘same’ reason also outweighs the ‘same’ other reason the next time too. In this connection different kinds of a fortiori arguments are possible too. For example, if in the first case there was only one reason why some reason set outweighs another reasons set, in a next case there may even be an additional reason why this is so. Then there are good grounds to assume that also in the next case the first set of reasons outweighs the second set.
Value judgments do not only deal with the question whether something is good, beautiful, or polite, but also with issues such as which act is better than which other act, or what is the best level to regulate criminal procedure. Such judgments are comparative value judgments and their justification (or the justification of the agent in accepting them) is somewhat more complicated than the justification of ‘ordinary’ value judgments. As a matter of fact, we already dealt to some extent with comparative value judgments, since the balancing of sets of reasons is an example of something that at least strongly resembles comparative valuation, since the judgment that one reason set outweighs another set is comparable to the judgment that the former reason set is ‘better’ than the latter.
Just like with the comparison of reason sets, the easy case for comparative value judgments is when the items which must be compared derive their values from the same scale. One level of regulation is better than another level if the only relevant aspect is the efficiency of the regulations and if the one level is more efficient than the other. If two rules must be compared and they both have only one good making characteristic, namely that they increase the level of employment, the rule that increases the level of employment more is the better rule.
As these two examples illustrate, comparative value judgments can sometimes be based on characteristics which obtain to a larger or lesser degree. The promotion of employment is bigger or smaller; the efficiency of a level of regulation is bigger or smaller; one picture was painted craftier than another and is for that reason better.
However, this easy mode of comparison is often not available because the one item has a different good or bad making characteristic than the other. Incommensurability lurks again. The ‘solution’ is the same as with balancing reason sets: the determination that one alternative is better than another may merely be a matter of decision making. Once such a decision exists, it functions as a precedent for future comparative value judgments. For instance, if it is was once decided that the efficiency of regulating contract law of the EU level is more important than the autonomy that is promoted by regulating contract law on the level of national states, efficiency will normally be more important than autonomy in a next case too. And, again, a fortiori arguments are possible here too. If one alternative has been found better than another in the past, a similar new alternative that has the good making characteristic to an even bigger degree while its competitor is the same as its predecessor, the new alternative will almost certainly be better than its competitor. For instance, if in a new case the gain in efficiency is even bigger than in the old case in which the efficiency was already decided to be more important than the autonomy, efficiency will a fortiori win out in the new case.
Some value judgments are not concerned with the question what is the better item, but what is the best item. For instance, what is the best level to regulate contract law? These absolute comparative value judgments can be justified by identifying a number of alternatives and by showing that the allegedly best item is better than all of the alternatives in the set. In this connection the argument can be made easier by assuming that the better-than relation is transitive: if item A is better than B and B is better than C, then A is also better than C. There may be exceptions to transitivity, but then the burden of proof is on who wants to make the exception.
The topic ‘burden of proof’ leads us to the final approach to the justification of value judgments, the procedural one.
The word ‘justified’ suffers from the process/product ambiguity. It may stand for the ‘rightness’ of a judgment such as a value judgment or of an agent accepting a judgment or other position. It may also stand for a finished process of justification. If a judge argues in court why she will condemn a suspect to three months imprisonment, she is involved in the procedure of justifying this verdict. After she has successfully finished this procedure, the verdict has been justified. That the verdict was justified in this procedural sense does not tell us anything about the substantial quality of the judgment, about its correctness or legal or moral qualities. It only tells us that the procedure of justification has successfully been brought to an end. However, if this procedure can only be completed successfully if the verdict satisfies certain substantial conditions, there is a connection between the verdict having been procedurally justified and its content being substantially justified (correct).
In this section we will have a closer look at the justification of value judgments by means of a procedure. To have a better understanding of procedural justification in general, we will start with a distinction between three kinds of procedures aimed at gathering ‘information’ in a very broad sense, including information about what is good, better, or best.
To acquire information it is always necessary to follow some procedure. This procedure may involve sensory perception, such as when we look to see something, or listen to hear something. It may also involve reasoning, such as when we multiply 3 and 5 to know the value of the product of these two numbers, or when we hypothetically fill in numbers in order to solve a Sudoku puzzle. More serious examples are the reasoning performed by a judge in order to ‘find’ the correct outcome of a case, or of a legal scientist who tries to figure out whether contract law is better regulated at the level of national states or at the EU level.
Because the acquisition of information always requires the application of some procedure, it is important to know how the outcome of a procedure relates to the correctness of the information provided by the procedure. In this connection Rawls has made a useful threefold distinction, on the basis of two questions. The first question is whether the correctness of the information provided by the procedure depends on the procedure. A lottery is an example where this is the case. What the winning ticket is depends only on the correct application of the rules of the lottery; there is no independent standard to determine which ticket has won. It will for instance not do to argue that the wrong person was pointed out by the drawing, because somebody else needed the money more. Moreover, there is no reason to assume that repeated application of the procedure should lead to the same outcome. Procedures like this, where the proper outcome is defined in terms of a correctly executed procedure, are called ‘pure procedures’.
The proper amount of income tax that a citizen should pay is also determined by means of a procedure. In the Netherlands, for instance, the amount of tax to be paid is decided by the tax inspector. However, there are tax rules that determine, at least for standard cases, which amount of tax the inspector should come up with. The procedure of having the tax inspector establish the tax assessment leads, if it is executed perfectly, to the proper amount of taxes. Procedures like this, where the proper outcome is defined independent of any procedure, but where some procedure, if executed well, is guaranteed to lead to the correct outcome, are called ‘perfect procedures’. If a perfect procedure is correctly executed more than once, it will always produce the same correct outcome.
There are also procedures that aim to reach an independently existing proper outcome, but where there is no guarantee that the procedure, even if executed perfectly, leads to the correct outcome. A jury trial aimed at determining whether the defendant did what he was accused of is an example of such a procedure. The proper outcome, conviction if and only if the defendant did what he was accused of, is determined independent from the procedure, and the procedure is designed to reach the proper outcome as often as possible, but there is no guarantee that a properly executed criminal trial always leads to the correct outcome. Such procedures, where the proper outcome is defined independent of any procedure, but where the procedure is not guaranteed to lead to the correct outcome, are called ‘imperfect procedures’. There is no guarantee that repeated execution of an imperfect procedure leads to the same outcome every time.
In our discussion of coherentism we have seen that the theoretical ideal of constructing a coherent position set is unrealistic. Human beings are unable to turn their initial incoherent position set into a coherent one. As a potential remedy we considered defeasible coherentism, according to which an agent tries to construct a coherent representative subset of its full position set and determines on the basis of this subset whether a particular value judgment is justified. The feasibility of this approach depends on the possibility to construct a representative subset and the problem in this connection is that it will not be possible to compare the subset with the full set for which it should be representative. A perfect substantive test whether the subset is representative is therefore not possible.
However, it is possible to work with an imperfect procedure. If an agent has sufficient reason to assume that a consistent subset of his full position set leads to a different outcome concerning a particular value judgment than the coherent reconstruction of his full position set, then the agent should not go by the subset in its present state, but revise it until he has no sufficient reason anymore to assume that the subset is not representative. Then the agent is justified in believing that the subset is representative and in accepting the value judgment on the basis of this subset.
The procedure in this connection is to keep modifying the position subset until it is consistent and the agent has no reason anymore to believe that the subset is not representative. This procedure is imperfect because there is no guarantee that repeated execution of this revision procedure all times leads to the same outcome.
An altogether different procedural approach to the justification of value judgments takes dialogs as its starting point. The basic idea is that justification is justification for a particular audience. A value judgment counts as justified if …, and then there are two possibilities:
- the audience has actually accepted the value judgment as result of the procedure;
- the value judgment logically follows from what the audience has accepted in the course of the procedure.
In the discussion of syllogistic/deductive justification we have seen that the main problem lies in the justification of the premises of a justificatory argument. As soon as the premises are given, the remainder is a matter of logic: does the value judgment logically follow from the accepted premises or not? In first instance (section 5.2) in particular the choice of a standard as starting point for the evaluation appeared to be problematic, since standards are not true or false. However, it turned out that judgments of fact fare little better than standards, since it is not truth or falsity that counts, but acceptance being justified. Acceptance of facts is in this connection not really different from acceptance of standards (section 6.4).
How can procedural justification help to overcome the above foundational problem? By replacing justified premises by premises that were actually accepted in a dialog between the proponent of the value judgment and her audience. Instead of an extensive specification of how that works, the following example may for the present purposes give sufficient insight.
This was a good soccer match!
Do you think so?
The audience does not accept the value judgment without justification.
Yes, the play was offensive, the players are technically gifted and there was hardly any foul play.
That is all correct, but does it make the match into a good one?
The proponent mentions the facts on which his value judgment is based.
The audience accepts these facts, and these are now settled for the dialog.
However, the audience has still doubts about the standard that was (implicitly) used.
What more should we expect from a good soccer match?
Well, that Liverpool wins … ;-)
The proponent attempts a shift in the burden of proof: if the audience does not agree with the standard, it should provide (and if necessary justify) a better standard itself.
Apparently the audience accepts this shift in the burden of proof, by ‘proposing’ an alternative standard.
You cannot be serious …
No you are right. Actually your standard is okay.
The proponent casts doubt on the standard proposed by the audience.
The audience does not even attempt a justification and accepts the standard originally proposed by the proponent.
So you agree that the match was a good one.
Well, if you insist: yes.
In the end the audience accepts the value judgment which has now been justified. Obviously, the audience could have made the proponent’s job much harder by not accepting the shift in the burden of proof and by asking for justifications for everything the proponent claims. If the audience clings to that strategy, the proponent will ultimately loose the dialog because he has the burden of proof.
This is an example of a dialog with rules that give the participants plenty of room. They can, for instance, make fake claims, or they can try to change the dialog rules in the course of the game (shift the burden of proof). The example dialog is purely procedural. Any judgment proposed by the proponent and accepted by the audience counts as justified, no matter whether it is ‘really’ true. It is also possible to have dialogs which are imperfect procedures that aim to lead only to conclusions that are actually true or justified. That can, for instance, be promoted by disallowing participants in a dialog to accept conclusions that do not follow logically from what has previously been established, or by committing the participants to a set of rules, standards and factual beliefs about which they have no choice (e.g. the rules of positive law and facts that are generally known). As a matter of fact, there are many variants possible.
It is possible to define many different kinds of procedures and depending on their definitions, these procedures will be pure, perfect, or imperfect. However, they have in common that their outcomes count as justified because they are the outcomes of the procedures. In a sense, judicial procedures as they are used in law are an example of procedural justification. That holds even for value judgments given by judges. If judges are bound by legal rules or standards, judicial procedures are examples of perfect or imperfect procedures. If judges exercise a discretionary power, they are examples of pure procedures.
In the previous sections several variants and aspects of the justification of value judgments have been discussed. Perhaps the main conclusion was that there is nothing special about the justification of value judgments. The fact that their justification will be based on standards which lack a truth value does not raise the complications that are traditionally connected to this use of standards. Not that the complications do not arise; they do. The reason why the justification of value judgments does not differ fundamentally from the justification of other judgments is that the complications arise with all judgments, and not in particular in connection with value judgments. The complications are the result from the impossibility to find an unshakeable foundation on which value or other judgments can be built. There is no such a foundation and the foundationalist approach to justification is doomed to fail.
There exists an alternative for foundationalism, however, and that is coherentism. A judgment is justified, or – better - an agent is justified in accepting a judgment, if this judgment is an element of a coherent position set held by this agent. What coherence amounts to is a topic that deserves an extensive discussion. Above it was only very briefly argued that a position set is coherent if and only if it is comprehensive, consistent and complete. There is no procedure that guarantees to lead to such a coherent position set, and neither is there a guarantee that an agent can only arrive at one coherent position set. This is less problematic than might seem at first sight, since the issue at stake is not any more whether a value judgment is justified, but whether an agent is justified in accepting a value judgment.
Because a coherent position set will normally be infinitely large, there is a guarantee that no human agent will ever arrive at a coherent position set. The alternative for full-blown coherentism is to be satisfied with defeasible coherentism: an agent is justified in accepting a judgment if and only if this judgment is an element of a consistent position set of which the agent is justified in believing that it is with regard to this judgment representative for the full coherent position set at which the agent would ideally arrive.
In this connection the development in time of an actually held position set into a coherent position set is an example of a procedure in which the agent is both the proponent and the opponent of all the positions in his position set. The rules in his position set determine how this procedure of updating his position set can correctly take place in time.
Similar rules can guide procedures in which two or more agents develop their ideas about a particular judgment. If such a procedure leads to agreement on a particular judgment, the proponents of this judgment have justified the judgment for the audience consisting of the opponents of this judgment. Noteworthy in this connection is that a third notion of ‘being justified’ is at stake. It is not anymore the value judgment as such that is justified, nor the agent who is justified in accepting the value judgment. In the procedural approach, the value judgment has been (rather than is) justified before a particular audience.
In this last section, an attempt will be made to translate the findings of the previous sections into a brief method for setting up an argument that leads to a value judgment about the best level of regulation for European private law. In this connection it is most fruitful to treat the justification of judgments, including all value judgments and all legal judgments, as a procedure in which the proponent of the judgment tries to make his audience accept the judgment that he wants to justify. If this procedure is conducted in a rational fashion, the proponent adduces reasons to convince the audience. The audience counts as being convinced if either it actually accepts the judgment, or is rationally committed to accepting it, e.g. because it follows logically from what the audience has in fact accepted.
The simplest argument is to give a decisive reason for the judgment. Such a reason is a fact that is relevant for accepting the judgment and which makes all counterarguments superfluous. A deductively valid argument would provide such a decisive reason, on the assumption that all the premises are true. The demand that all the premises are true can be weakened to the demand that the audience accepts all the premises. Given that the argument is deductively valid, an audience that accepts al the premises is rationally committed to accepting the conclusion. For example, if the audience agrees that European private law should be created as close to the people as possible and that regulation by national governments is as close to the people as possible, then the conclusion must be that European private law should be created by national governments.
If an audience does not accept one or more of the premises of an argument, these premises can be treated as judgments standing in need of justification, and what is recommended here as general procedure to justify judgments can be applied to these premises too. So, if the audience does not believe that regulation by national governments is as close to the people as possible, then an additional argument should be presented that has this thesis as its conclusion.
Most often a judgment cannot be justified by means of a decisive reason, and then the best possible rational approach is to justify it by means of contributory reasons. The proponent of a judgment has the burden to adduce one or more contributory reasons for the judgment. Any such reason has two components which need to be made explicit: an alleged fact, and a standard that declares the fact to be supportive for the judgment. Both the fact and the standard need to be accepted by the audience, and if that is not immediately the case, the proponent has the burden of proof for both the fact and the standard. The same procedure that applies to all judgments also applies for this justification.
For instance, one might argue that European private law should where possible be created by private agents, and present as a reason for this conclusion that the preferences of agents are best satisfied by regulations made by the agents themselves (alleged fact), a reason that is made relevant by the standard that it is the purpose of regulations to satisfy the preferences of the persons to whom these standards apply.
If the audience has accepted one or more contributory reasons for the judgment, it is rationally committed to accepting this judgment, unless it produces counter-reasons. The audience has then the burden of proof for these counter-reasons. Assume for example that an audience accepts that the fact that the preferences of agents are best satisfied by regulations made by the agents themselves is a reason for letting the agents make the regulations themselves. If this audience is not convinced, it has the burden to prove a counter reason. Such a counter reason might for instance be that regulation by the national government prevents a race to the bottom.
The same procedure that holds for pro-reasons also holds for counter-reasons. If the audience succeeded in making the proponent accept one or more counter-reasons, the proponent has the burden to convince the audience that the contributory reasons that plead for the judgment outweigh the contributory reasons that plead against it. In this connection the proponent must explicitly indicate what the pro- and con-reasons are, and consequently what the required ‘weight-premise’ involves. Only if the proponent succeeds in making the audience accept this ‘weight premise’, the audience is rationally committed to accepting the judgment.
For instance, the proponent of regulation by the agents must convince the audience that the reason based on the satisfaction of personal preferences outweighs the reason of the race to the bottom.
If an argument is presented on paper, for instance in order to convince the readership of a scientific journal, the author takes both the roles of the proponent and of the audience. The author must therefore not only offer reasons of which he assumes that they will convince the readership, but must also critically ask whether these reasons are really convincing, and whether there are (in the eyes of the readership) no reasons pleading against his conclusion. The quality of such an argument depends on how good the author judges his actual readership, and on how serious she takes her task to represent a critical audience in justifying the judgment that she defends.
Aarnio, A., The Rational as Reasonable. A Treatise on Legal Justification, Dordrecht: Reidel 1987.
Albert, H., Traktat über kritische Vernunft, Tübingen: Mohr 1980.
Alexy, R., Theorie der juristischen Argumentation (7e Auflage), Frankfurt a/M: Suhrkamp 1983.
Alvarez, M., Kinds of Reasons. An Essay in the Philosophy of Action, Oxford: Oxford University Press 2010.
Araszkiewicz & Šavelka 2013
Araszkiewicz, M. & Šavelka, J. (eds.), Coherence: Insights from Philosophy, Jurisprudence and Artificial Intelligence, Dordrecht: Springer 2013.
Ayer, A.J., Language, Truth and Logic, Harmondsworth: Penguin 1971 (2nd edition; 1st edition 1936).
Bagnoli, C., ‘Constructivism in Metaethics’, in: E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2011 edition). URL = <http://plato.stanford.edu/ archives/win2011/entries/constructivism-metaethics/>.
Baldwin, T., ‘The Open Question Argument’, in J. Skorupski (ed.), The Routledge Companion to Ethics, Abingdon: Routledge 2010, p. 286-296.
Blackburn, S., Essays in Quasi-Realism, Oxford: Oxford University Press 1993.
Canaris & Larenz 2014
Canaris, C-W & Larenz, K., Methodenlehre der Rechtswissenschaft (4e Auflage), Berlin: Springer 2014.
Chang, R. (ed.), Incommensurability, Imcomparability, and Practical Reason, Cambridge: Harvard University Press 1997.
Dahlman & Feteris 2012
Dahlman, C. & Feteris, E. (eds.), Legal Argumentation Theory: Cross-Disciplinary Perspectives, Dordrecht: Springer 2012.
Dworkin, R., Taking Rights Seriously (2nd edition), London: Duckworth 1978.
Dworkin, R., Law’s Empire, London: Fontana 1986.
Feigl & Scriven 1956
Feigl, H. & Scriven, M. (eds.), Minnesota Studies in the Philosophy of Science (volume I), Minneapolis: University Of Minnesota Press 1956.
Fuller, L.L., The Morality of Law (revised edition), New Haven: Yale University Press 1963.
Geach, P.T., ‘Good and Evil’, 17 Analysis, 1956, p. 33-42.
Ginsberg, M.L. (ed.), Readings in Nonmonotonic Logic, Los Altos: Morgan Kaufmann Publishers 1980.
Golding, M.P., Legal Reasoning, Petersborough: Broadview 2001.
Gordon, Th.F., The Pleadings Game. An Artificial Intelligence Model of procedural Justice, Dordrecht: Kluwer 1995.
Haack, S., Philosophy of Logics, Cambridge: Cambridge University Press 1978.
Haack, S., Evidence and Inquiry. A Pragmatist Reconstruction of Epistemology (2nd edition), Amherst: Prometheus Books 2009.
Hage, J.C., Reasoning with Rules, Dordrecht: Kluwer 1997.
Hage, J.C., Studies in Legal Logic, Dordrecht: Springer 2005.
Hage, J.C., ‘The Method of a Truly Normative Legal Science’, in: M. van Hoecke (ed.), Methodologies of Legal Research, Oxford: Hart Publishing 2011, p. 19-44.
Hage, J.C., ‘Construction or Reconstruction? On the Function of Argumentation in the Law’, in: C. Dahlman & E. Feteris, Legal Argumentation Theory: Cross-Disciplinary Perspectives, Dordrecht: Springer 2012, p. 125-144.
Hage, J.C., ‘Legal Reasoning and the Construction of Law’. 7(16) i-Lex, 2012, p. 81-105 (<http://www.i-lex.it/us/previous-issues/volume-7/issue-16/103-legal-reaso ning-and-the-construction-of-law.html>.
Hage, J.C., ‘Three Kinds of Coherentism’, in: M. Araszkiewicz & J. Šavelka, Coherence: Insights from Philosophy, Jurisprudence and Artificial Intelligence, Dordrecht: Springer 2013, p. 1-32.
Hage, J.C., ‘The Deontic Furniture of the World’, in: J. Stelmach, B. Broźek & M. Hohol, The Normativity of Law, Kraków: Copernicus Press 2011, p. 73-114.
Hare, R.M., The Language of Morals, Oxford: Oxford University Press 1952.
Hare, R.M., Freedom and Reason, Oxford: Oxford University Press 1963.
Haslett, D.W., ‘Idealism, philosophical’, in: T. Honderich (ed.), The Oxford Companion to Philosophy, Oxford: Oxford University Press 1995, p. 860 please mention first and last page of article It is an encyclopedia lemma and completely on one page
Hempel, C.G., Philosophy of Natural Science, Englewood Cliffs: Prentice Hall 1966.
Honderich, T. (ed.), The Oxford Companion to Philosophy, Oxford: Oxford University Press 1995.
Jones, O.R., ‘Supervenience‘, in: T. Honderich (ed.), The Oxford Companion to Philosophy, Oxford: Oxford University Press 1995, p. please mention first and last page of article. It is an encyclopedia lemma and completely on one page
Lodder, A., Dialaw. On Legal Justification and Dialogical Models of Argumentation, Dordrecht: Kluwer 1999.
MaCormick, N., Legal Reasoning and Legal Theory, Oxford: University Press 1978.
MacCormick, N., Rhetoric and the Rule of Law, Oxford: University Press 2005.
Mill, J.S., ‘Utilitarianism’, in: M. Warnock (ed.), Utilitarianism, London: Collins 1962, p. 251-321.
Moore, G.E., Principia Ethica, Cambridge: Cambridge University Press 1903.
Peczenik, A., On Law and Reason (2nd edition), Dordrecht: Springer 2008.
Pollock & Cruz 1999
Pollock, J.L. & Cruz, J., Contemporary Theories of Knowledge (2nd edition), Lanham: Rowman and Littlefield 1999.
Rawls, J., A Theory of Justice, Oxford: Oxford University Press 1972.
Raz, J., Practical Reason and Norms, London: Hutchinson 1975.
Redondo, C., Reasons for Action and the Law, Dordrecht: Kluwer 1999.
Reiter, R., ‘A Logic for Default Reasoning’, in: M.L. Ginsberg (ed.), Readings in Nonmonotonic Logic, Los Altos: Morgan Kaufmann Publishers 1980, p. 68-93.
Searle, J., Speech Acts, Cambridge: Cambridge University Press 1969.
Sellars, W., ‘Empiricism and the Philosophy of Mind’, in: H. Feigl & M. Scriven, Minnesota Studies in the Philosophy of Science (volume I), Minneapolis: University Of Minnesota Press 1956, p. 253-329.
Skorupski, J. (ed.), The Routledge Companion to Ethics, Abingdon: Routledge 2010.
Smits, J.M., The Mind and Method of the Legal Academic, Cheltenham, Edward Elgar 2012.
Sosa, E., ‘The Raft and the Pyramid: Coherence versus Foundations in the Theory of Knowledge’, 5 Midwest Studies in Philosophy, 1980, p. 3-25.
Stelmach, Brożek & Hohol 2013
Stelmach, J., Brożek, B. & Hohol, M. (eds.), The Normativity of Law, Kraków: Copernicus Press 2011.
Stevenson, C., Ethics and Language, New Haven: Yale University Press 1944.
Toulmin, S., The Uses of Argument, Cambridge: Cambridge University Press 1958.
Van Eemeren & Grootendorst 2000
Van Eemeren, F.H. & Grootendorst, R., Kritische discussie, Boom: Amsterdam 2000.
Warnock, M. (ed.), Utilitarianism, London: Collins 1962.
Wiarda, G.J., Drie typen van rechtsvinding (3rd edition), Zwolle: Tjeenk Willink 1988.
 Italics added in all three examples.
 See Wiarda 1988, p. 23, Alexy 1983, p. 23, and MacCormick 2005, p. 73/4.
 Smits 2012, p. 35 and 41.
 Hage 2011.
 For references, see the other contributions to this volume.
 For instance, Aarnio 1987, Alexy 1983, Canaris & Larenz 2014, Dworkin 1986, MacCormick 1978, and Peczenik 2008.
 This deficit is remedied to some extent in Hage 2011, 2013a and 2013b.
 For examples from several fields, see Hempel 1966, p. 49-54, Hare 1952, p. 32-55, Alexy 1983, p. 273-283, and MacCormick 1978, p. 19-52.
 This argument can be found in Moore 1903, chapter I.
 Baldwin 2010.
 An example of something that does depend for its existence on minds is a sound. Sound is the way in which longitudinal vibrations in the air are experienced. If there were no minds, there would not be experiences of sound either, but only vibrations,
 Philosophers who dare to do so used to be called ‘idealists´ or ‘non-realists’, although the term ‘constructivists’ becomes popular too. See Haslett 1995 and Bagnoli 2011.
 Representatives of this view in meta-ethics are Ayer 1936, Stevenson 1944, Hare 1952 and Blackburn 1993.
 Geach 1956 and Searle 1969.
 Because value judgments are on this view true or false, it can be argued that the view is cognitivist, be it that the cognition concerns facts that depend on argument and therefore on the mind.
 According to Bagnoli 2011, ‘constructivism in ethics is the view that insofar as there are normative truths, …, they are in some sense determined by an idealized process of rational deliberation, choice or agreement’.
 Dworkin 1986, p. 225.
 Smits 2012, p. 76.
 Extensive discussions of constructivism in law can be found in Hage 2012a and 2012b.
 The point that value judgments are supervenient has been emphasized in Hare 1963, chapter 2, be it that Hare used the expression ‘descriptive meaning’ for what is presently called ‘supervenience’.
 Jones 1995, defines ‘supervenience’ as a dependency relation according to which one set of properties (the supervening properties) is so related to a second set (the subvening properties) that there could be no difference in the first without there being a difference in the second, although there could be a difference in the subvening properties without a difference in the supervening properties.
 It is assumed that the author’s needs concerning monitors did not change.
 The development of case law for this particular subject is nicely discussed from a logical point of view in Golding 2001.
 The presence of liability supervenes on characteristics of a case, but this does not imply that being liable is a value judgment. Value judgments are supervenient, but so are many other judgments.
 The word ‘rule’ is between quotes, because the rule is the outcome of universalization, and not – as is customary in a legal setting - the product of rule creation.
 For some value judgments other facts than about these characteristics may be relevant too. To simplify the argument, this complication will be ignored.
 See, for instance, Haack 1978, p. 13-15.
 The field dependence of standards (‘warrants’) is one of the main points of Toulmin 1958.
 Logicians should notice that this gives the standard two roles which are traditionally considered to be distinct. In the first place the standard provides the ‘major premise’ needed for the justificatory argument. In the second place it provides the inference rule that guarantees that the result of applying the standard to a true description of the facts leads to a justified judgment. In terminology that used to be fashionable, the standard is a ‘material inference rule’ as opposed to the formal inference rules that are used in traditional logics. Reiter’s (1980) ‘defaults’ are another example of such material inference rules.
 Some might argue that in the common law legal reasoning by analogy is based on identifying the ratio decidendi of a precedent, which is quite similar to identifying a standard by means of universalization. That there exists a similarity is correct, but that is a reason to interpret ‘analogizing’ on the basis of a ratio decidendi as a kind of rule-based reasoning, where the rule is derived from a precedent (rather than a statute or other regulation), and not as reasoning by analogy properly speaking.
 In section 6.4 it will be argued that this move, which seems to solve the problems, is in the end not satisfactory.
 Cases in point are the views of Alexy 1983, p. 273-283 and of MacCormick 1978, chapter II.
 We ignore here the complication that reasoning on the basis of premises that lack a truth value cannot be deductively valid, and that some other standard for the quality of arguments needs to be applied.
 This argument is based on Fuller’s (1963) account of the internal morality of law.
 This is compatible with the phenomenon that the application of such standards still requires the exercise of judgment, which may lead to differences between evaluators.
 This possibility, in combination with the view that legal principles are not identified on the basis of their pedigree, underlies Dworkin’s famous argument in Dworkin 1978, p. 14-45.
 Mill 1861.
 Albert 1980, p. 13.
 Sellars 1956.
 An example of an attempt can be found in Peczenik 2008, p. 132-140.
 I will return to this impossibility in section 7.3.
 There are even more possibilities, such as making an exception to a standard which is maintained for other cases. Discussion of these possibilities is outside the scope of this article, however.
 Haack 2009, p. 65/6.
 Sosa 1980.
 A position is defeasible if it is possible that it needs to be modified in the light of additional information. A legal example would be that the conclusion that a buyer of a car needs to pay the price of the car, is liable to be defeated in the light of the information that the sales contract came about under undue influence. More on defeasibility in connection with law in Hage 2005, p. 7-32.
 More extensive accounts can be found in Hage 1997, chapter IV, and in Hage 2005, p. 69-99.
 A good general account of reasons that guide behaviour is Alvarez 2010. An account that focuses on legal reasons for action is Redondo 1999.
 The notion of an exclusionary reason was popularised in a legal context by Raz. See, for instance, Raz 1975, p. 40-48. In epistemology, the same phenomenon has become known as an ‘undercutter’ or an ‘undercutting defeater’. See Pollock and Cruz 1999, p. 196.
 The distinction has a legal counterpart in the distinction between void and a rescinded contract.
 See Chang 1997.
 ‘same’ is between quotes because reasons are facts and two cases will seldom consist of the same facts. The facts may be of the same kind. For details see Hage 2005, p. 101-134.
 Details in Hage 2005, p. 101-134.
 A different example of an a fortiori argument will be given in section 9.
 Details and more examples can be found in Hage 2005, p. 101-134.
 Rawls 1972, p. 84-86.
 An exception would be that relevant facts have changed in between the execution of the procedures.
 The expression ‘sufficient reason’ is used for the situation that either there is a decisive reason, or that one set of reasons outweighs another set.
 Notice that this demand is weaker than the demand that the agent has reason to assume that the subset is representative.
 That there is no such guarantee has to do with the fact that there is a test to recognize a coherent set, but no procedure for arriving at such a set. This means that it is at least theoretically possible to follow different routes to a coherent position set, and also that these different routes lead to different outcomes, all coherent. That would be problematic if the function of a coherent position set were to represent an independently existing reality. However, that is not its function. Its function is to indicate when an agent is justified in holding the position set he actually holds, and it is very well possible that an agent might be justified in holding different position sets, although not at the same time.
 Examples of elaborated sets of dialogue rules can be found in Gordon 1995 and in Lodder 1999. A general discussion of different kinds of procedural justification is Hage 2005, p. 227-264.
 This is more or less the starting point of the ‘pragma-dialectical’ approach to argumentation. See Van Eemeren & Grootendorst 2000, p. 57-61.