Jusletter IT

When asked about a solution to a complicated legal problem, lawyers will often state that «it depends on the facts of the (concrete) case». This contention, although intuitively understandable to almost all professional participants to legal discourse, leads to important theoretical problems. The phrase quoted above is typically invoked in cases involving the application of vague, open-textured or evaluatively open legal rules. But the configuration of facts is also extremely important in case-based reasoning, in particular when there is no obvious leading precedent at hand. The collision of legal principles (for instance, in constitutional review) is another context in which the circumstances of the case may play a decisive role. One configuration of facts may lead to ascription of priority to one of competing constitutional values and another configuration may support the opposite conclusion. Note that the acceptance of one interpretation of a vague rule and not another one or creation of a newratio decidendi obviously affects legally relevant values. Therefore, in each of the three abovementioned contexts there is an interplay between facts of the case, its outcome and legally relevant values. The present paper intends to focus exactly on this interplay and to provide a framework for this area of legal argumentation.

More specifically, I will concentrate on the three following research questions:

1. What is the structure of value judgments employed in legal balancing?
   
   
2. What are the relations (in particular: possible collisions) between factor-based arguments and teleological arguments in legal balancing?
   
   
3. What are the functions of factor-based reasoning in the context of teleological reasoning in the law?

Obviously it makes sense to talk that facts have influence on the possible outcome of the case only if we take into consideration the normative links between certain facts and outcomes and not the facts as such.¹

These issues have been the point of interest of legal scholars since many decades. In the 19th century and in the first half of the 20th century the problem discussed here was the focus of the so-calledInteressenjurisprudenz and of many scholars in American jurisprudence, including Holmes and Pound (cf. [2] for references). In the 70s, 80s and 90s of the 20th century the problem of legal balancing in concrete cases was extensively debated by prominent theorists of legal argumentation like Alexy [3], MacCormick [4] and Peczenik [5]. During the recent two decades the relation between facts, values and legal reasoning has become one of the main points of interest of researchers involved in formal analysis of legal reasoning, especially in the field of Artificial Intelligence and Law. For instance, one should mention the work of Ashley [6], Bench-Capon and Sartor [7], Bench-Capon and Prakken [8], Grabmair and Ashley [9], Hage [10] Prakken and Sartor [11], Rissland and Ashley [12] and Sartor [13]. Many problems concerning the structure and rationality of case-based reasoning and teleological reasoning have been quite well explained in the abovementioned research. It is not the case that we have abundance of mutually incompatible conceptions which eventually turn to be useless in argumentative practice. The contrary claim holds: the formal work on the frameworks of reasoning discussed here brought many insightful contributions to legal theory, and particularly to the field of AI & legal argumentation. For instance, it is obvious now that legal reasoning involving values possesses much more complicated structure than simple legal syllogism and it seems that it is probably best accounted for by «theory construction» approach [7]. But this does not mean that the problem of legal balancing is not challenging anymore. In particular, although it is not very difficult to enumerate different elements or aspects of legal balancing (for example: rule-based reasoning, factor-based reasoning and teleological reasoning) it is not always easy to grasp the mutual relations between these aspects. To quote Peczenik: «A weighing a balancing ofprima facie reasons is a jump: one has premises, the conclusion and a gap between. The inference is not conclusive. One performs a weighing of reasons, in spite of not having a precise knowledge as to how much they do «weigh». In legal argumentation, such jump occurs in different contexts. Not all of them concern outright weighing of values. Sometimes it is difficult to tell what is weighed, or if there is a weighing involved in a jump. [14]» The late legal philosopher from Lund captured very briefly the following intuitions: (1) not all weighing in legal reasoning involves weighing of values; (2) there are cases in which it is difficult to classify the weighed object; (3) there are cases in which, although we expect the weighing to occur, the reasoning eventually applied involves no weighing at all. This paper aims to account for these difficulties analytically. The order of investigations is as follows. First, I try to classify the types of value judgments involved in the process of legal balancing, basing on the old distinction between classificatory and comparative value judgments. The second point of analysis is the distinction between values and factors in legal reasoning. The framework developed here aims to support the thesis that factors may lead judges to conclusions of legal argumentation without actually engaging in the process of legal balancing. However, sometimes it is necessary to formulate all the relevant value judgments explicitly. In such cases, factor-based reasoning enters into different relations with weighing of values. I try to elucidate the structure of these interdependencies. The next part of the paper is devoted to argumentation moves employed in legal balancing. The most important aim of this part of article is to analyze the possible conflicts among arguments based on values and arguments based on factors. It is checked how these types of conflict fit into well-known classifications, for instance into the distinction between rebutting collisions and undercutting collisions. Finally, I indicate some problems for further research.

Before we begin analyzing the structure of value judgments, let us focus on the concept of value. It is convenient to treat values as criteria of evaluation, applicable to certain states of affairs. I do not postulate any particular ontology of values (and of states of affairs) here; for the sake of our analysis it is sufficient to adopt a ‘juridical’ conception of values, according to which values are criteria of evaluation (1) identified as (legally) relevant by at least considerable part of the lawyer’s community and (what seems to be the other side of the same coin) (2) extractable from what is considered the so-called sources of law (let us note that this need not amount to criteria directly mentioned in the legislative text; our account is broader and more general). The typical examples of legally relevant values are liberty, freedom of information, privacy, public security and so on. Although it is very difficult to provide for precise definition of a value, it is important to note that values may be promoted (realized) or demoted (infringed) (1) by certain behavior patterns performed by parties of legal relations and, perhaps even more importantly, (2) by certain judicial decisions. The conceptual link between values and the notion of promotion/infringement is crucial here, for not all criteria of evaluation function in such way. For instance, a car may be assessed from the point of view of its speed, but it seems very unnatural to state that the «value of speed» is promoted by purchase of this car. The judicial decision-making aspect of balancing is discussed in the literature (for instance [3], [6], [9], [13], particularly in the context of judicial application of the so-called principles (understood as optimization requirements by Alexy [3], [15]) or goal-norms (recently: Sartor [2]). For the sake of simplicity I will not attempt to conceptually distinguish values from goals. The optimization commands of goal-norms demand a given value (a goal) be realized to the greatest extent possible, given legal and factual possibilities. Importantly, the legal possibilities are set both by legal rules (which may operate as exclusionary reasons constraining teleological reasoning (recent account in [2])) and by colliding legal values (goals). In the latter case it is necessary to establish the relative weight of colliding values. Due to the fact that all legally relevant values cannot be maximized simultaneously, legal decision-makers have to balance them against each other for the sake of finding an optimal solution. This is the most important context for teleological reasoning in the law, taking place in judicial application of law, in judicial review and in legislation. Involvement in teleological reasoning entails formulation of different types of value judgments.

According to Alexy it is possible to distinguish classificatory, comparative and metric value judgments [3]. Classificatory value judgment consists in stating that a state of affairs is good, bad or neutral from the point of view of a given value. Comparative value judgments introduce the relation of preference: one state of affairs may be preferred to another one in the light of the degree of realization of a given value. Finally, metric value judgments assign numbers to the degrees of realization of different values by states of affairs. Accepting this classification as a point of departure, let us analyze the abovementioned types of value judgments in a more systematic way.

Let S be the set of all states of affairs (s). Let the symbol V+(s) have the following meaning: a state of affairs s promotes value V. The symbol V- (s) would, then, mean that a state of affairs infringes value V. The two symbols here represent opposite classificatory value judgments. It is possible to express a neutral classificatory value judgment in terms of the two already defined ones. We can state that a state of affairs s is neutral in respect of

The intuitions behind classificatory value judgments seem quite obvious at the first sight. For instance, if an issue of a daily is banned due to the fact it has published some texts which make the government feel uncomfortable, then, one would say, the ban obviously infringes the value of freedom of information. But even these simple value judgments seem to be context-sensitive. For instance, in Poland in 1989, when communism begun to collapse, the parliamentary elections were organized in the following way. The elections to the Senate were free, but the communist government was entitled to appoint a significant number of representatives to the Lower Chamber. Should this situation be assessed as good or bad from the point of view of realization of value of democracy? From the point of view of Western democratic standards such regulation of elections is obviously bad, but actually it was assessed as good in comparison to earlier times in Poland. Similarly, in the 18th century hardly anyone thought in Europe that corporal punishment generally infringes human dignity. To conclude: even though classificatory value judgments have relatively uncomplicated structure, they can nevertheless appear problematic in certain contexts and therefore their soundness may be contested in legal argumentation.

Comparative value judgments play a pivotal role in legal balancing. They enable us to express the intuition that different states of affairs may realize (or infringe) values to different degrees. This can be represented by a kind of superiority or preference relation. The expression «Veff (Sx) > Veff (Sy)» means that the state of affairs Sx is better than Sy with respect to the effect of this state of affairs concerning value V (that is, either Sx promotes V to the greater extent than Sy or Sy diminishes V to the greater extent than Sx). We can refer to this expression as to simple comparative value judgment.

This issue is particularly important in the contexts in which two or more values are relevant and these values are mutually incompatible,i.e. they support different outcomes. In such cases it is necessary to perform a compound comparative value judgment, related to the overall effect of the set Sx on the set of relevant values VR = {V1, V2, …Vn} and the overall effect of the set Sy on the very same set of relevant values: «VReff (Sx) > VReff (Sy)». To establish these two overall effects we need to introduce the concept of degree of (non)satisfaction of value and to assume that these degrees are comparable. The exact account of what is compared with what in this context is somewhat controversial in the literature and it is not my aim to discuss this issue thoroughly here. It is sufficient to recall the four following proposals. I attempt to formulate a condition for a given choosing one of the competing outcomes rather than another one according to each of the accounts presented below. I will refer to them as Solution Choice Conditions (SCC).

• Classical proposal (Alexy). According to the famous «Law of Balancing» the degree of non-satisfaction of one set of colliding values should be compared to the importance of realization of another set of values [3]. Alexy introduced additional variables in a structure referred to as «The Weight Formula» [15]. SCC: according to this approach (in a simplified version based on Law of Balancing only), a solution Sx (required by one of the sets of competing values) ought to be preferred rather than Sy if and only if the degree of importance of realization of the set of values supporting Sx, is bigger than (it outweighs) the scope of interference with the opposing set of values, supporting Sy.
  
• Constraint satisfaction approach. Weighing of values in the law may be accounted for by computing coherence of mutually constrained elements, in particular of legally relevant values and legal conclusions. What is compared here is the relative weight of coherence and incoherence relations between values and outcomes [16]. SCC: a solution Sx ought to be preferred rather than Sy if and only if Sx belongs to the set of accepted elements generated after dividing the initial set of elements so as to maximize coherence,i.e. the total weight of all satisfied constraints.
  
• Weighing values in hypothetical reasoning framework. According to Grabmair and Ashley [9], a value judgment is defined as the comparison of value effect sets, namely, the sets of (value, effect) tuples of values positively, negatively and neutrally affected by imposition of a given fact on a given situation []. Therefore, the authors identify the concept of value judgment with what we refer to as compound comparative value judgment. SCC: a solution Sx ought to be preferred rather than Sy if and only if the value effects set involving Sx outweighs the value effect set involving Sy.
  
• Marginal analysis. Sartor postulates to compare differential benefits of one considered choice against another one with respect to a given set of values. In order to obtain a value judgment here one need to count the sum of multiplications of realization of relevant values by the weights given values have in intervals at issue [2]. SCC: a solution Sx is superior to Sy with regard to a given set of values if and only if the mentioned sum is above zero. This account refines and develops Alexy’s proposal concerning representation of legal balancing with indifference curves and the decreasing marginal benefit thesis (compare [2] and [3]).

The application of different types of metrics to the abovementioned proposals automatically transforms comparative values judgments into metric value judgments. It is not my aim here to analyze these proposals thoroughly or to investigate into the problem whether they are mutually translatable and under which assumptions. This would involve another, perhaps much longer, study. However, it seems interesting to point out that the gradual hierarchy of different types of value judgments discussed here, with respect to the degree of complication of their structure, strictly resembles the hierarchy employed in classical theory of scales of measurement as advanced by Stevens [17]. Classificatory values judgments are strictly analogical to nominal scales a simple comparative value judgments to ordinal scales. Application of different metric types leads to formulation of metric value judgments which resemble interval or ratio scales. Although this analogy is not very useful for the practice of legal argumentation, it may be effectively employed in theoretical (mathematical) models of legal reasoning.

The presentation of different types of values judgments employed in legal argumentation leads to the conclusion that successful teleological reasoning involves much information, indeed, often much more than it is accessible for legal decision-maker. I will return to this problem in what follows by presenting different types of conflicts of value-based arguments. At this point I shall argue that the problem with teleological reasoning in the law consists not only in the limited capacities of human mind and difficult access to (the existing) data, but also in the fact that some of the relevant information has to be developed by a legal decision-maker in the process of legal reasoning. My argument goes as follows:

• Let us assume that the aim of law is to optimize the realization of legally relevant values. This does not mean that all legal reasoning should be teleological because due to some factual constraints sometimes non-teleological reasoning brings bigger benefits to legally relevant values than explicit balancing of values.
  
• In consequence, if parties to legal relations behave in such a way that they yield a net benefit to legally relevant values, legal decision-makers should not intervene.
  
• If the conduct of legal actors leads to net loss in realization of legally relevant values, legal decision-makers ought to intervene in order to correct the loss generated by the parties (this is the main idea of corrective justice).
  
• However, in order to perform a corrective justice exercise it is necessary to first know how the distributive justice issues are settled. These latter issues will be often the subject of the dispute between the parties. Both parties may agree that there should be no net losses in realization of legally relevant values, but one party may be convinced that her action did not lead to such loss while the other one can claim to the contrary. The legal decision-maker has to decide which of the parties’ claims is justified and assess its own potential decisions from the point of view of distributive justice.

Although there may be many sources of disagreement in argumentation between the parties to the dispute, I would like to indicate the following one: both parties can maintain that the SCC is satisfied by the outcome desired by a given party. The teleological reasoning structure is therefore insufficient to provide an outcome to the case unless we can determine which of the conflicting solutions satisfies the SCC. Due to the fact that human rationality is bounded, the solution to the case may be hardly accessible [2]. More importantly, the solution may be underdetermined in the following sense: the set of beliefs concerning values relevant for the SCC-classification of competing solutions may be constructed along two or more incompatible manners with no clear choice criterion between these options [18]. These considerations bring some authors to the conclusion that values should not be seen as an autonomous source of practical determinations ([13] p. 221).

The concept of weighing of values may be easily generalized to the notion of balancing of all pro- and con-reasons for (against) a given conclusion [5], [10]. Among different types of reasons the concept of factor is of special interest here. According to a definition provided by Sartor, factors are circumstances which prompt our physical or mental behavior in certain directions [13]. Factors can be best accounted for as generalized features of cases. Importantly, these features are legally relevant, because they are linked to potential solutions of the case. It is possible to distinguish the so-called binary factors (which either obtain or not obtain in a given case) and scalable factors (dimensions), which can manifest in different degrees in cases. That a certain feature of a case is accounted for by a binary or a scalable factor is to some extent a matter of convention. For instance, it is possible to state that presenting false information is a binary factor in a case involving a journalist and a politician: a journalist either presents false information or not. But it is also possible to state that falseness of information is a dimension: the more false information concerning the politician has been published in a newspaper, the more it is likely that the journalist will lose the case. Consequently, if the publication contains no false information at all, this circumstance favors the journalist. Therefore we can see that sometimes it is useful to divide the dimension into two parts thereby changing it into binary factors [13]. It is worth noting here that the concept of dimension had been introduced earlier than the concept of factor. Rissland and Ashley present a comprehensive insight concerning the relations between the two concepts [12].

Factors may support or not support potential solutions to a legal case. If they are scalable, they can do this to different degrees. The problem of additive character of factors is disputable but it is often claimed that either accrual or merging of factors is possible (see [13] p. 234 ff. for the discussion). Let us now return to the nature of the link between factors and solutions. As it was stated above, this link represent propensity so it is weaker than the relation between the antecedent and a consequent of a rule. Therefore, even if one is able to identify a factor of the case which favors a given solution, it does not follow automatically that this solution ought to be accepted. However, sometimes the concept of «rule» is used to refer to factor/outcome links [7].

Although both factors and values may be weighed, they should be distinguished from each other. Factors and values play are differently related to outcomes or conclusions of reasoning: while values can be promoted or infringed by adoption of a given conclusion, factors are elements of the pre-existing context which may favor or disfavor certain outcomes ([13], p. 178]. In other words: preferring of a given legal solution and not another one may lead to realization (of infringement) of certain values, given certain factors are present (or absent) in description of the case at hand. This pint may be captured by the following reasoning scheme (here referred to as Factor-Outcome-Value relation):

For instance, if my aim is to promote the value of good relations with friends, and if paying a visit to my friend because it is his birthday endorses these good relations, then there is a reason for visiting a friend on his birthday. Let us note that this kind teleological justification is not necessary to establish the factor-outcome link. The factor [friend’s-birthday] prompts the outcome [visit] even if teleological justification if this link is not explicit. And this is one of the most important functions of factor-based reasoning: they constitute a simplification of teleological reasoning. Let us assume that we have to decide between two incompatible solutions: Sx and Sy. In order to justify a choice between the two teleologically, one should (1) identify all relevant values affected by choosing Sx or Sy; (2) determine the marginal benefits arising from realization of each of the relevant values in Sx and Sy and to compare these differential benefits (here we are employing the SCS as advanced by Sartor [2010] but it is possible to adopt other SCS accounts, too). Obviously, this may be a difficult task even at the first move, concerning identifying the relevant values. However, factor-based reasoning is not so demanding. The only thing a decision-maker is supposed to do is to check whether in the description of the case there exist factors prompting one or another solution. Comparison of relative weight of sets of factors supporting mutually incompatible conclusions may be quite difficult after all, but after all it does involve the degree of complication typical for teleological reasoning.

Let us illustrate this point with the following simple example. Assume that Thomas is a professor of law and he is considering what to do during the weekend: on the one hand, he can go skiing, on the other hand, he can stay home. There are several factors involved in the case. The weather forecast for weekend is very good and reliable (which favors skiing) and a good friend asked Thomas to go skiing together (another pro-factor). But the deadline for paper submission for an important conference is also set for this weekend and moreover Thomas has recently had some problems with his knee (these two factors favor staying at home). Factor-based reasoning involves comparison of the two mentioned sets. On the other hand, teleological reasoning would force Thomas to identification of all relevant values here (health, friendship, scientific advance) and to engage into complicated marginal analysis. The consideration of relevant importance of pro- and con-factors (for instance, by conceding that the problems with knee are too serious to risk a ski trip in the nearest future and in consequence choosing staying at home) seem to be much more useful at this point, although it does not guarantee that the choice made was theright one.

It seems that factor-based reasoning has been present in legal literature concerning legal balancing ever since, although it was not presented under this heading and the structure of factor-outcome links was not accounted for analytically. Recall the classical discussion of theLebach case as discussed by Alexy [3]. The facts of this case were as follows: (1) television channel ZDF intended to broadcast a documentary about a serious crime concerning murder and stealing of weapons; (2) one of the people involved in this crime (a minor offender) was about to be released from prison at the time of intended broadcast; (3) the authors of the program intended to publish the name and a picture of this person. The prisoner applied to the court for injunction and demanded that the program should not be published, mainly because it would endanger his resocialization. The German Federal Constitutional Court engaged in teleological reasoning and indentified two relevant conflicting values: the value of personality and freedom of press. Then the Court stated identifying the circumstances justifying giving precedence to one of these values. First, the Court determined that freedom of press takes precedence before the protection of personality rights if media engage in ‘up-to-date reporting of crime’. However, the Court did not identify this circumstance in the description of the case. Instead, four factors supporting together the protection of personality were indicated, which resulted in adoption of the following rule: «a repeated (F1) media report, no longer required by the interest of current information (F2), concerning a serious criminal offence (F3), which endangers the resocialization of the offender (F4) is constitutionally prohibited» ([3] p. 56).

The description of the Federal Constitutional Court’s reasoning as presented by Alexy resembles the structure of factor-based reasoning rather than fully-fledged teleological reasoning as accounted for above. Although the values at stake are identified, the rest of the reasoning is essentially factor-based (with one qualification pertaining to the nature of the link between the circumstances of the case and the outcome, see below). The Court did not focus much on factors supporting the conclusion in favor of the ZDF, but concluded with stating that a potentially media-supportive factor (current information) is not present in the description of the case. The emphasis of the opinion was on the factors of repetitive character of the broadcast and on the endangerment to the resocialization process (accounted for as binary and not scalable factors, which introduces even more simplicity to the reasoning). It should be emphasized at this point that according to theoretical account of balancing, as advanced by Alexy, if a given set of circumstances yields a certain preference relation between competing values, and the prevailing value is favored by a certain solution, then a valid rule applies with the set of circumstances as its antecedent and the solution (which promotes the prevailing value) as a consequent. Therefore, the priority relation between the values plays a role of a ‘middle term’ in this structure of reasoning. The preference ordering between the values may be, then, eliminated from the explicit structure of argumentation, as it can be done in case of factor-based reasoning, even if it is ultimately teologically justified. The main difference between classcal approach to the subject (as defended by Alexy) and factor-based approach is consists in the structure of the link between the (set of) circumstances of the case and its outcome. According to legal philosopher from Kiel, this link can be accounted for with the concept of conditional rule, while factor-outcome links are much weaker (although this link may be contextually sufficient). This difference stems from the more general assumptions present in Alexy’s theory of legal argumentation, which cannot be discussed here. Let us note that the difference mentioned here may not be as serious as it seems, because in the theory of factors it is admissible to transform a certain (perhaps, well established) factor-outcome links into (in)defeasible rules. In consequence, it may be stated that Alexy anticipated much of the research on factor-based reasoning, although he did not make use of this concept.

The discussion presented here clearly indicates that factor-based reasoning provides an alternative approach to (fully fledged) teleological reasoning in legal balancing. The relation between values can be simply accounted for as «priority relation» and it may function as a middle term in reasoning, while main focus is on the factors (dis)favoring certain outcomes. Yet, the discussion presented above (in particular of theLebach case) leads to some important questions concerning the relation of factors and values in legal balancing. It seems obvious that the Federal Constitutional Court’s opinion, as expressed in the Lebach case, even though it may be correct after all, could be attacked on different grounds. Therefore, the relation between factor-based reasoning and teleological reasoning in legal argumentation should be accounted for in more systematic way. Due to the fact that there are many important contributions to this field [6–13] I will attempt to focus on argumentative schemes which have not been discussed very thoroughly so far.

It is not my aim here to provide a rigorous formal account of argumentation framework concerning factors and values. I suggest that the schemes discussed below can be incorporated into well-developed formal frameworks like these advanced by Hage [10], Prakken and Sartor [11] or Bench-Capon and Sartor [7]. Although I consider the latter framework particularly fit for the problems presented below, for simplicity of the present account I will follow rather the footsteps of Prakken as presented in [19] and concentrate on conflict between arguments and not on comparison between whole theories. The comparison of arguments presented here should be seen as a preliminary proposal and a point of departure for a richer framework designated to comparison of competing theories of cases. As for SCC, I follow the recent Sartor’s proposal [2] concerning marginal analysis of conflicting values (although the adoption of other interpretation of SCC is also possible). I also adopt FOV account as discussed above, noting that this is not the only possible view on this relation. For the sake of simplicity let us assume that in discussed cases there are exactly two conflicting values: Vi and Vj, promoted by the incompatible legal solutions (conclusions, outcomes) Sx and Sy respectively. I employ the following general argumentation scheme: (1) a factor-based argument in favor of a given solution is presented by a proponent (referred to as Arg), (2) a counterargument based on values judgment is stated by an opponent (Con), and then (3) the possible reactions of the proponent of the first argument are considered (Re) .

I employ the following notation:

• [Factor-outcome link] F ⋇ Sx represents the factor-outcome link such that a factor F,
  
  present in the description of the case at hand, favors adoption of solution Sx for this case;
  
  
• [Factor-outcome links priority] Fi ⋇ Sx > Fj ⋇ Sy holds if and only if the set of factors
  
  favoring Sx is stronger than the set of factors favoring Sy.
  
  
• [Positive Classificatory Value Judgment] Vi+(Sx(F)) represents the positive classificatory value judgment in respect of value Vi concerning adoption of solution Sx regarding the presence of factor F;
  
  
• [Negative Classificatory Value Judgment] Vi-(Sx(F)) represents the positive classificatory value judgment in respect of value Vi concerning adoption of solution Sx regarding the presence of factor F;
  
  
• [Degree of Realization] D (Vi+(Sx(F))) represents the degree of realization of value Vi concerning adoption of solution Sx regarding the presence of factor F;
  
  
• [Degree of Dissatisfaction] D (Vi-(Sx(F))) represents the degree of dissatisfaction of value Vi concerning adoption of solution Sx regarding the presence of factor F;
  
  
• [Difference] Δ (Vi) (Dx, Dy) represents the difference in realization of value Vi between choosing solution Sx and solution Sy.
  
  
• [Weight] w (Vi) (Sx, Sy) represents the weight of interval represented by Δ (Vi) (Dx, Dy), as regards the value Vi. (The latter four definitions are necessary to account for the marginal analysis SCC conceptual framework).
  
  
• [Comparative Value Judgment] ((Δ (Vi) (Dx, Dy))* (w (Vi) (Sx, Sy)) + (Δ (Vj) (Dx, Dy))* (w (Vj) (Sx, Sy))) > (Δ (Vj) (Dy, Dx))* (w (Vj) (Sy, Sx)) + (Δ (Vi) (Dy, Dx))* (w (Vi) (Sy, Sx)) represents the statement that the sum of marginal benefit as regarding Vi and marginal loss as regarding Vj, given Sx is chosen is greater than the sum of marginal benefit as regarding Vj and marginal loss as regarding Vi, given Sy. The benefits are always represented in positive, while losses – in negative numbers. Obviously, it is possible to formulate a comparative value judgment favoring Sy and a comparative value judgment concerning that the two solutions are indifferent as regards the values at stake. For the sake of readability we will abbreviate the lengthy formula to «CVJ». The CVJ form adopted here stems from the SCC interpretation as presented in [2].

Let us start our presentation with argumentation concerning classificatory value judgments solely. The scheme for such argumentation would go as follows.

[Argument Collision Scheme 1 .]
[Factor-based argument attacked by a classificatory value judgment]

1. Arg.: F ⋇ Sx , therefore Sx.
   
2. Con: (1) Sx iff Vi+(Sx(F)), but
   
   (2) ∼ Vi+(Sx(F)). Therefore:
   (3) ∼ Sx; therefore Sy.

Let us note that the type of counterargument presented here may have two distinct interpretations, concerning the justification of premise Con (2), namely:

I do not want to focus here excessively in the possible reaction of the proponent. It seems sufficient to state than he can engage in either negating Con (1) (thus introducing some new factors favoring his outcome) or in negating Con (2) which can consist primarily in precedent-based reasoning.

The argument scheme concerning comparative value judgment is a much more complicated structure. It is assumed that at the case at hand the problems of classificatory value judgments are not controversial. For the sake of simplicity we will ignore the possibility of equal differences of weights, because this would lead us to the discussion of decision-making in the context ofnon sequitur , namely, to the problem of burden of argumentation and there is no space here for this discussion.

[Argument Collision Scheme 2.]
[Factor-based argument attacked by a comparative value judgment]

1. Arg.:Fi ⋇ Sx > Fj ⋇ Sy, therefore Sx.
   
2. Con:

The array of possible reactions of the proponent is very broad. Let us note that the Con (2) (a–d) arguments need justification, too, and it is natural to state that this justification can be built upon factor-based reasoning. This is a very important point: factors can be linked not only directly with the outcomes, but also with the statements concerning the relations between differences of realization of values and between the weights of intervals in question. The proponent may try to undercut the Con (2) (a – d) arguments, but it is more interesting to look at the potential rebutting reactions. This kind of reasoning will consist in removing the (ceteris paribus) clause from the arguments of the opponent. Let us note a list of such rebutting reactions for the argument Con (2) a.

1. Re: (1) although Con (2) (a) holds, either:

However, such way of responding to the counterarguments is difficult and very demanding from the cognitive point of view. This illustrates how the switch from factor-based to (fully-fledged) teleological reasoning makes it considerably complicated. Therefore, it seems suitable to look at the possible reactions which would enable the proponent, attacked by an comparative value judgment arguments go back to factor-based reasoning. These reactions can be accounted for as follows(they should be seen as possible continuations of Argumentation Collision Scheme 2. Let us present three such reactions.

[Factor Clustering and Rule Formation]

In this argumentation move the proponent claims that the factors of the case are so strongly linked to a given outcome that this outcome should be adopted, given the factors are present. The adopted rule would presumably have defeasible character, but the courts may not be eager to adopt another outcome in presence of the defined set of factors, unless extremely important reasons point to the contrary.

[Argument from the Core Threshold]

The concept of the core threshold of rights and values is a well-known issue in theory of constitutional rights [3] and it is invoked in legal-theoretical discussions concerning teleology [2]. The core threshold is the minimal degree of realization of value which generally cannot be infringed in any case decided. Thus, instead of engaging in comparative teleological reasoning it may be sufficient for the proponent to indicate that the adoption of solution defended by the opponent leads to the beyond-core-threshold infringement of the value promoted by the solution defended by the proponent.

[Lack of Necessary Condition]

The proponent may refrain from engaging into teleological discourse by pointing out that the solution defended by the opponent can be accepted only if certain factor is present in the description of the case and that this factor is not present. It seems that the Federal Constitutional Tribunal used this heuristic when discussing the absence of ‘interest of current information’ factor.

The discussion presented above leads us to the formulation of a thesis concerning the function of factors in legal balancing. Factor play the three following functions with regard to teleological reasoning in law: (1) approximating function, (2) constraining function and (3) determining function. The first function is the most obvious one. Reasoning with factors provides shortcuts which enable the decision-maker to adopt a solution which (arguably) promotes the values at stake in optimal manner. Therefore, factor-based legal decisions may be seen as approximations of teleologically optimal decisions. As for the second function, it was apparent from our discussion that the factors may constrain teleological reasoning by pointing out certain outcomes and preclude the decision-maker from further balancing. This function can be also seen as indirect approximation of teleological optimality, for the constraining factors may favor such outcomes which generally lead to good results as far as optimization of values is concerned (although in certain situations they may detract from it and bring Pareto-suboptimal results). Finally, factor-based reasoning may serve not only as an approximation, but as an output for determination of optimal balancing result in a given case – if the background legal knowledge (doctrine, precedents,) does not provide sufficient hints for the established optimal balancing. In such cases it is necessary to analyze the case at hand from the point of view of broader theories (in particular, theories of values) which should be compared with each other. It should be noted that while the first and the second function serve as tools for realization of corrective justice, the third function involves distributive justice considerations as well. As far as value judgments-based reasoning is concerned, it may serve as a correctional device for factor-based reasoning, if the latter is oversimplified.

In the present contribution I classified the types of value judgments employed in legal reasoning and investigated how factor-based reasoning interacts with arguments based on value judgments. The preliminary results, concerning argumentation relevant argumentation schemes and three functions of factor-based reasoning with regard to legal balancing (and the link of this problem to corrective and distributive justice idea) seem to offer a promising point of departure for further research. I would like to indicate the three following areas of this research.

First, the framework outlined above should be incorporated into a more rigorous and well-developed formal account of legal reasoning. This can enable the researches to assess the validity of presented argument schemes in conclusive manner. It seems necessary to enrich the framework to focus more extensively on scalable factors (dimensions) and hierarchies of factors.

Second, it seems necessary to analyze a large body of case material (including the decisions of Court of Justice of the European Union, European Court of Human Rights and constitutional courts, including the US Supreme Court) to investigate how the argumentation schemes discussed above are employed in practice. Such research would probably lead to the identification of numerous argument moves which are not mentioned here, and in particular, of constraining factor-based moves.

Third, the argument moves discussed here should be analyzed from the perspective of broader theories (as advanced in Bench Capon and Sartor). This context creates a perspective for integration of the presented outcomes with constraint satisfaction theory of coherence [20].

1. MacCormick N., Weinberger O.: An Institutional Theory of Law. D. Reidel Publishing Company, Dordrecht/Boston/Lancaster/Tokio (1986).

2. Sartor G.: Doing justice to rights and values: teleological reasoning and proportionality. Artificial Intelligence and Law 18, 175–215 (2010).

3. Alexy R.: A Theory of Constitutional Rights. Transl. J. Rivers. Oxford University Press, Oxford (2002) (1st German ed. 1985).

4. MacCormick N.: Legal Reasoning and Legal Theory. Clarendon Press, Oxford (1978).

5. Peczenik A.: On Law and Reason. Springer (2008) (1st ed. 1989).

6. Ashley K.D. : Modelling Legal Argument. Reasoning with Cases and Hypotheticals. MIT Press, Cambridge, Mass. (1990).

7. Bench-Capon T., Sartor G.: A Model of Legal Reasoning with Cases Incorporating Theories and Values. Artificial Intelligence 150, 97–142 (2003).

8. Bench-Capon T., Prakken H.: Using argument schemes for hypothetical reasoning in law. Artificial Intelligence and Law 18, 153–174 (2010).

9. Grabmair M., Ashley K.D.: Argumentation with Value Judgments. An Example of Hypothetical Reasoning. In: Winkels R.C. (ed.) Proceedings of the Twenty-Third Annual Conference on Legal Knowledge and Information Systems (JURIX). IOS, Amsterdam, 67–76 (2010).

10. Hage J.C.: Studies in Legal Logic. Springer, Berlin (2005).

11. Prakken H., Sartor G.: Modelling Reasoning with Precedent in a Formal Dialogue Game. Artificial Intelligence and Law 6, 231–287 (1998).

12. Rissland E.L., Ashley K.D.: A note on dimensions and factors. Artificial Intelligence and Law 10, 65–77 (2002).

13. Sartor G.: Legal Reasoning. A Cognitive Approach to the Law. In: Pattaro E. (ed.) A Treatise of Legal Philosophy and General Jurisprudence. Vol. 5. Springer, Dordrecht (2005).

14. Peczenik A.: Jumps and Logic in the Law. What Can One Expect form Logical Models of Legal Argumentation?. Artificial Intelligence and Law 4, 297–329 (1996).

15. Alexy R.: On Balancing and Subsumption. A Structural Comparison. Ratio Juris 16, 433–449 (2003).

16. Araszkiewicz M.: Balancing of Legal Principles and Constraint Satisfaction. In: Winkels R.C. (ed.) Proceedings of the Twenty-Third Annual Conference on Legal Knowledge and Information Systems (JURIX). IOS, Amsterdam, 7–16 (2010).

17. Stevens S.S.: On the Theory of Scales of Measurement. Science 103, 677–680 (1946).

18. Peczenik A., Wróblewski J.: Fuzziness and transformation: Towards explaining legal reasoning. Theoria LI, 24–44 (1985).

19. Prakken H.: An exercise in formalizing teleological case-based reasoning. Artificial Intelligence and Law 10, 113–133 (2002).

20. Thagard P.: Coherence in Thought and Action, MIT Press, Cambridge: Massachusets, London, England (2000).

---

Michał Araszkiewicz is an adjunct lecturer in the Department of Legal Theory at the Faculty of Law and Administration of the Jagiellonian University in Cracow.

Jagiellonian University, Department of Legal Theory, Faculty of Law and Administration, Bracka St. 12, 31-005 Cracow, Poland
michal.araszkiewicz@uj.edu.pl

---