A. The Trouble with Distinguishing

According to Lamond, this position cannot be squared with the role of distinguishing in the practice of reasoning by precedent.Footnote ³² The problem lies in the kind of rule the straightforward version of the rule-approach presupposes. If a rule is supposed to fully determine every case that falls under its antecedent, then the inference between premises and conclusion in the employed legal syllogism needs to be certainty preserving (in that it warrants us to be as sure of the conclusion as we are of the premises) and monotonic (it does not lose any strength if additional premises are added to the argument). The strength of the argument should not change if we move from argument (a) to argument (b).

But, as Lamond points out, this is exactly what happens with precedent. The practice of distinguishing allows judges not to reach decision Y on the basis that the present-case has further property n5—if n5 is not also present in the precedent-case.Footnote ³³ Consider the case of distinguishing in MacDonald v. Ortho Pharmaceutical Corp. Footnote ³⁴ Carole MacDonald suffered a stroke caused by her taking Ortho-Novum birth control pills. Ortho had not provided a specific warning of strokes, and MacDonald sued under “Failure to Warn Doctrine.” The court acknowledged that the case fell under a narrow exception to the general duty to warn. This exception had been set out in the Restatement (Second) of Torts in a canonical, rule-like formulation and subsequently adopted by courts.Footnote ³⁵ It stated that a manufacturer has no duty to warn a consumer if warnings have been given to a responsible intermediary. Cases adopting this rule had determined that prescribing and treating physicians were responsible intermediaries. However, even though the court agreed that the case fulfilled the conditions set out in the rule-like formulation, it distinguished because of an additional property of the case: unlike other medication, where doctors are the main decision-makers, patients usually make the decision to take birth control pills relatively autonomously.

Lamond argues that it is indeed an authoritative reason for following if a present-case exhibits the properties listed in the ratio of a precedent-case, but it is not necessarily a binding reason.Footnote ³⁶ To be a binding reason, it must be an authoritative reason that excludes all other reasons, except maybe for specifically listed exceptions.Footnote ³⁷ Judges would need to have the duty to follow precedent whenever their present-case falls under the antecedent of the precedent-rule. But instead, judges have the duty to either follow or distinguish. A legal syllogism that uses a rule formulated from a precedent's ratio cannot be certainty preserving. Even if a present-case exhibits the properties listed in the rule, and the interpretation of the precedent is correct, the application of the rule's consequent might not follow.

For this reason, Lamond proposes to think of precedents differently. Instead of excluding all other reasons from consideration, a precedent only excludes those that were already present in the precedent-case.Footnote ³⁸ A reason not to follow the precedent based on the property n5 would be excluded only if n5 was already present in the precedent-case. Otherwise, the judge must take the reason into account and can distinguish on its basis.Footnote ³⁹ Lamond's argument shows that interpreting the precedent's opinion to extract a rule and determining whether the resulting legal syllogism is valid is not enough. To determine that a precedent gives a binding authoritative reason to follow, the judge must make sure that there are no nonexcluded reasons against following based on additional properties of the present-case.

It is possible to amend the rule-approach to accommodate Lamond's criticism. This is so because Lamond is interested in the kinds of reasons precedents provide and aims at showing that they are not the same as those provided by statutes. By contrast, this paper is concerned with the kind of reasoning judges can employ to recognize those reasons. Lamond allows for opinions to provide a ratio in the form of a conditional. He points out that this conditional cannot be used like a statute because a precedent can be distinguished based on unenumerated reasons. To incorporate this criticism, the rule-approach can make use of a defeasible modus ponens. The concept of a defeasible modus ponens comes from the flourishing project of developing nonmonotonic logics—logics that are equipped to deal with premise-conclusion relations that are less than certainty preserving but that, given that the premises are true, still establish a presumption in favor of the conclusion.Footnote ⁴⁰ This presumption can be defeated if other reasons enter the consideration. Authors like Horty (with Bench-Capon) and Rigoni have developed approaches to reasoning by precedent that use defeasible rules and a nonmonotonic logic based on Lamond's insights.Footnote ⁴¹ In their (amended) rule-approach of reasoning by precedent, the reasons capable of defeating the presumption for following are ones against following based on aspects that the present-case has but that the precedent-case did not have. It should be clear that amending the rule-approach in this way means reading the “rule” that the precedent establishes as an extremely defeasible rule that is much closer to a rule of thumb (though it excludes some reasons from considerations) than some proponents of rule-approaches would like.Footnote ⁴² The argument-scheme below illustrates the change that results from moving from Alexander and Sherwin's strict approach to Lamond's, Horty's, Bench-Capon's, and Rigoni's defeasible model:

According to one reading of the amended rule-approach, the additional critical question imports an element of the analogy-approach. Under this reading, the question requires judges to determine whether any of the differences between the precedent-case and the present-case are legally relevant. Once the rule has been extracted from the precedent-case, carried over to the present-case, and applied there, the judge still has to bring both cases together in order to reason about their differences.

There is another possible reading of the amended rule-approach, according to which no direct comparison between the cases is necessary. According to this reading, interpreting the precedent-opinion would involve cataloging the properties that cannot serve as a basis for distinguishing because they are present in the precedent-case. If a judge wanted to distinguish based on some property of the present-case, she would have to consult this list to determine whether her reason for distinguishing was excluded.Footnote ⁴³ This would preserve the purity of the rule-approach, avoiding all direct contact between the two cases. I do not want to claim that Horty and Bench-Capon think this is actually how reasoning by precedent works, but I think that their account makes it plausible that it could work this way at least theoretically.Footnote ⁴⁴ Nonetheless, I am not sure that this actually happens in practice. In the next section, I will discuss the difficulties of determining a ratio that is exact enough to serve as the major premise in a modus ponens, defeasible or not. I think these difficulties would be even greater in an attempt of making a list of properties as suggested above.

B. The Trouble with Opinions

So far, I have been talking about the interpretation of judicial opinions as if it is a simple task to derive a ratio from an opinion and make a rule from it. Sometimes, this might be the case. MacDonald v. Ortho Pharmaceutical Corp made use of a formulation provided in the American Restatement (Second) of Torts. In the United States, restatements are written with the aim of summarizing the common law in canonical formulations. The results are paragraphs that look almost like statutes.

However, it turns out that retrieving a rule from an opinion comes with its own set of difficulties.Footnote ⁴⁵ A rule lists a number of relatively clearly circumscribed categories in its antecedent that a case has to fall under in order for the rule to apply. Statutes usually come in a canonical formulation meant to designate these categories clearly. Opinions document the arguments that the precedent judge used to justify her decision in the precedent-case. They often do not use canonical formulations of rules, and where the categories are unclear later judges might need to employ first-order reasoning to interpret the opinion.Footnote ⁴⁶

Opinions might fail to give well-described categories for the aspects of the precedent-case meant to be singled out as its legally important factors.Footnote ⁴⁷ As members of the American Legal Realism movement have pointed out, categories come at certain levels of abstraction, and for the application of a rule it is important which level of abstraction this is.Footnote ⁴⁸ But the precedent judge might not have determined how broad the categories should be under which the factors of her case fall.Footnote ⁴⁹ Sometimes, the category is unclear because the court simply neglects to give one. At other times, a court might expressedly consider itself unable to give clear categories. Take, for example, Dillon v. Legg, in which the court decided that a defendant who negligently killed a child with his car can be held liable for the mother's emotional trauma, if the defendant could foresee the emotional trauma. Here, the court declares that it “cannot now predetermine the defendant's obligation in every situation by a fixed category; no immutable rule can establish the extent of that obligation for every circumstance in the future.”Footnote ⁵⁰ Instead it gives deliberately vague guidelines, one of which is that the victim and the emotional sufferer must have a close relationship. The opinion specifies that it is the aspect that the mother was closely related to the child that was important here, but what kind of close relation are we talking about? Must it be a blood relation? Or could best friends fall under this category? Given that the relationship between mother and child is about as close as it can be, any following case with a relationship between the victim and the emotional sufferer faced the question of whether the category extends far enough.Footnote ⁵¹ When opinions are unclear in this way, the present judge needs to fill out the missing information to get a usable rule. How she resolves this issue will have important consequences. Take the question of whether Dillon v. Legg applies only to relations of blood: if the judge decides that it does, then she has an authoritative reason to follow wherever there is a blood relation, but none where there is not.

There is then a sense in which unclear opinions do not really provide rules.Footnote ⁵² Instead, they provide a sometimes confusing collection of partial rules and hints for their completion. Sometimes there are fully formulated and explained categories. But sometimes, there are only some categories, and then some hints toward other categories, and some aspects that might or might not be necessary to consider.

On this basis, Adam Rigoni argues that the rule-approach cannot show that precedents are as constraining as it might appear after all.Footnote ⁵³ Some theorists, like Alexander and Sherwin, bite the bullet and claim that precedents with vague opinions do not constrain at all.Footnote ⁵⁴ But this seems to be a dissatisfying and problematic conclusion. For if a precedent can only be binding if its opinion provides a specific rule with explicitly circumscribed categories, then a considerable number of cases could not be sources of binding reasons. Of course, this is exactly what Leiter calls the “core claim” of American Legal Realism—that the law is indeterminate and that judges decide cases more on the basis of their perception of the facts than on the basis of legal reasons.Footnote ⁵⁵ I think that this pessimistic conclusion about precedent-constraint is probably unavoidable if we adopt an understanding according to which a precedent only constrains if its existence fully determines the decision of the judge. In this case, my paper simply shows that precedent comes equally close to being constraining in the rule-approach and in the analogy-approach. However, I adopt an understanding according to which judicial decision-making is constrained by a precedent if its existence either excludes reasons that the judge would otherwise have taken into consideration or adds reasons into the judge's considerations that would otherwise not have existed. Constraint comes in degrees.Footnote ⁵⁶

If we accept this more permissive understanding, then we can adopt a solution proposed by Shecaira and amend the rule-approach of reasoning by precedent again so that it reflects the inclusion of first-order reasoning on the level of determining the precedent-rule.Footnote ⁵⁷ Faced with a precedent-opinion that does not provide a rule, the judge chooses one among those rules that are supported by the information given in the opinion. She does so according to first-order reasons about which of these rules is the best one—legally and/or morally. Of course, it is important to say that when she does this, she must make sure that the rule she chooses is not in conflict with any of the surrounding law. Still, this changes the source and strength of the precedent-constraint rather dramatically. If the opinion does not readily provide a rule, then this constraint comes from the content available in the precedent-opinion that restricts the rules between which the judge can choose.Footnote ⁵⁸ This change can be represented in the argument-scheme as follows:

We are now at the point where we can ask what is left of the rule-approach's attractiveness. To recall, the argument was that according to the certainty-preserving, strict version of the rule-approach, precedents have a strong constraining effect on judicial decision-making. They provide judges with rules that can be used in certainty-preserving legal syllogisms, thereby relieving them of the need to engage in first-order normative reasoning. However, the two amendments that made the rule-approach consistent with the actual practice of reasoning by precedent imported first-order reasoning back in at two places.

First, the judge might employ first-order reasoning to determine whether the rule she identified was indeed the most preferable one among those consistent with the information provided in the opinion. Second, the judge employs first-order reasoning to determine whether there was a property that the present-case has and the precedent-case did not, and whether this property was a relevant reason against following.

The double-amended rule-approach includes first-order normative reasoning as an important feature of reasoning by precedent. I do not think that, as a result, precedents appear as completely unconstraining. After all, judges must choose rules that include all the information the opinion does provide, and that are not precluded by the law outside the precedent. Further, reasons against following based on aspects present in the precedent-case are excluded. It seems that the amended view now achieves some balance between constraint and freedom. However, I hope to have shown that among the rule-approaches currently on the market, those that can survive the objections are also the ones that depict precedents as providing a moderate amount of constraint. It is now worthwhile to inquire how well the analogy-approach of reasoning by precedent does in comparison.

A. Analogical Mapping

Analogies consist of the comparison of two analogs, a target-analog about which something is said and a source-analog that is used to say it. Imagine, for example, that I was to claim that Dillon v. Legg should be followed if an accident was not witnessed by the mother of the victim, but instead a kindergarten teacher. I might justify this by saying “kindergarten teachers are like mothers.” In this example, “kindergarten teachers” is the target while “mothers” is the source. In an analogy-approach of reasoning by precedent, the precedent-case serves as the source-analog, while the present-case serves as the target-analog.

In everyday life, we seem to make confident judgments about degrees of similarity as well as the appropriateness of source-analogs. According to the multi-constraint theory of analogy, we make these evaluations based on how successfully we can perform an analogical mapping between the two objects. The topic of analogical mapping is very complex, and I will provide a simplified overview here—just enough to develop my line of reasoning. Basically, analogical mapping consists in the establishing of a set of systematic correspondences between the source and the target.Footnote ⁶⁴ For example, the analogy “kindergarten teachers are like mothers” might lead you to this mapping:

Of course, we could simply claim correspondences between anything. For example, instead of mapping “love their wards” onto “love their children,” we could just declare the correspondence between “love their children” and “being a professional.” However, according to the multi-constraint theory, we feel the pull of several different constraints when we perform analogical mappings. The success of the mapping can be determined by how successfully we can adhere to these constraints. First, we consider mappings more successful if the corresponding aspects share surface similarities. Surface similarities are for example perceptive similarities in color or shape or semantic similarities in the kinds of concepts we use when we think about the analogs.Footnote ⁶⁵ Second, we tend to adhere to structural-constraint s.Footnote ⁶⁶ Above, we mapped attributes, but relations can also be mapped. For example, we could try to say that kindergarten teachers are like mothers by mapping the relation that mothers care for their children onto kindergarten teachers.

In the proposition “Mothers care for their children,” call “mothers” and “their children” the fillers of the slots of the relation “x care for y.”Footnote ⁶⁷ Then it becomes apparent that mapping a relation also involves mapping its fillers.Footnote ⁶⁸ Constraints apply here. In the example, the two relations are mapped because of their surface similarity. But the kindergarten teacher/mother and children/ward correspondences are established because they each fill the same slots: x stands for both mother and kindergarten teacher and y stands for both children and wards. A mapping of a relation is more successful if it is one to one, that is, if every element in the source corresponds to one unique element in the target.Footnote ⁶⁹ Given that the slots of “x cares for y” are filled by two different fillers in “Mothers care for their children,” it would be a weakness of the mapping if this was not so in the target. If you tried to map “Mothers care for their children” onto kindergarten teachers, and kindergarten teachers only cared for themselves, then your mapping (x care for x) would not be successful. Instead of a similarity, you would have found a difference.

A further constraint becomes apparent when we think about mapping higher-order relations, that is, relations of relations.Footnote ⁷⁰ For example, we could try to map the relation that mothers care for their children because they love them onto kindergarten teachers.

Again, we have to map the fillers to map the higher-order relations. Here, the mapping is better if the fillers are mapped consistently.Footnote ⁷¹ This means that whenever a relation is mapped, the fillers are also mapped. In the example, your mapping will be more successful if your observations of kindergarten teachers and their wards allow you to always map “wards” onto “children,” in all relations where “children” serves as a filler.

In his paper “Reason Is Too Large,” Dan Hunter presents evidence that judges engage in analogical mapping under both the surface-constraint and the two structural-constraints when they decide cases.Footnote ⁷² The surface- and structural-constraints explain how similarities and differences between two cases become apparent: the more successfully a judge maps aspects that are part of her precedent-case onto her present-case while adhering to the constraints, the more confidently the resulting correspondence can be called a similarity. Failure points to a difference.

However, these constraints do not explain why precedent could be constraining for judicial decision-making. The question remains which similarities and which differences are relevant. If you think about the analogy between kindergarten teachers and mothers, you will easily see that depending on what you pick among the many features that mothers possess, some mappings will be successful while others will fail. So there are similarities and differences. Which ones are relevant? Posed like this, that is a silly question. It cannot be answered without more information. Relevance is a relational concept—something is relevant with respect to some goal, task, or question; nothing is just relevant.Footnote ⁷³

Not surprisingly, cognitive scientists have found that when we perform analogical mappings, we adhere to a further kind of constraint—a purpose-constraint.Footnote ⁷⁴ Usually, we are not just presented with analogies. Analogies are drawn for some purpose, for example to solve a problem or to argue for a claim. If this is the case, analogical mapping is performed as a basis for analogical reasoning—to draw an inference about the target-analog.

Analogical inferences are made through a process called “copy and substitution.”Footnote ⁷⁵ Assume we have established a consistent correspondence between mothers/kindergarten teachers and children/wards. We know that it is true that “mothers love their children.” Do kindergarten teachers love their wards? We might try to get an answer through our analogy to mothers. We copy over the relation, substitute the fillers with kindergarten teacher equivalents and come to the assumption that “kindergarten teachers love their wards.” Now, this inference is not certainty preserving.Footnote ⁷⁶ Even if we are entirely certain that mothers love their children, the mere performance of a copy and substitution does not warrant any certainty about the relationship of kindergarten teachers and their wards.

But that does not mean that the claim “kindergarten teachers love their wards” cannot be supported by the mapping between kindergarten teachers and mothers. There are inferential connections between our knowledge that “mothers love their children” and other things we know about mothers and children. For example, we might know that there is a causal connection between the fact that mothers love their children and the fact that mothers protect/feed/play with their children: performing all these actions is causally (though not necessarily) connected to making the love grow. That gives us guidance for what to look for in kindergarten teachers.Footnote ⁷⁷ If we can successfully find correspondences to these aspects in kindergarten teachers while adhering to the surface- and structural-constraints, then the resulting mapping gives some support to the inference that “kindergarten teachers love their wards.” The more successful mappings we can perform with respect to aspects about mothers that stand in inferential relations to the aspect that mothers love their children, the better supported the analogical inference becomes.Footnote ⁷⁸ The purpose to answer the question whether kindergarten teachers love their wards has led us to concentrate on specific mappings between kindergarten teachers and mothers. The similarities and differences revealed through these mappings are the relevant ones.

B. Analogical Arguments and the Purpose-Constraint

In his paper, Dan Hunter expresses his belief that the purpose-constraint guides judges to pick those precedents that will allow them to make their preferred decisions in their present-cases. He points out that this conclusion would work well with the observations made by American Legal Realists, who claim that judges are guided more by their inclinations of which outcomes would be just than by legal reasons.Footnote ⁷⁹ This interpretation is in line with Frederick Schauer's criticism of the analogy-approach. Schauer believes that cognitive theory about analogical mapping shows that analogical reasoning cannot be constrained. Like Hunter, he thinks that the purpose-constraint guides judges to choose those source-analogs that allow them to make their independently favored decisions.Footnote ⁸⁰ I agree that analogical reasoning can give self-serving judges the option to wilfully determine how they would like to decide a case and then reject precedents because they cannot map their preferred decision onto the precedent-decision. However, I do not think that any method of reasoning could prevent such self-serving judges from rationalizing their favored decisions. This includes reasoning according the rule-approach, which makes rationalization possible because it necessitates the interpretation of opinions. The more interesting question is what analogical reasoning can do for judges who are honestly trying to evaluate the considerations both for and against following a specific precedent. Such judges do not try to find the right precedent for an argument that supports a conclusion they like. Rather, they try to determine whether a specific precedent can serve in a successful argument for a conclusion they might or might not like. In other words: they do not get to choose either source-analog or conclusion, but instead must evaluate whether a given source-analog can successfully support a given conclusion about a given target-analog. I will show that for judges in this position, the purpose-constraint of analogical reasoning is the key to understanding how precedents can exert constraint in the analogy-approach of reasoning by precedent.

Above, I said that the argument-scheme for the analogy-approach of reasoning by precedent contains an analogical argument in which the analogy supports the conclusion that the two cases are legally the same.Footnote ⁸¹ The inference to this conclusion is what a judge needs to evaluate to determine whether the precedent needs to be followed. The question becomes: Does the analogy with the precedent as the source-analog and the present-case as the target-analog support the conclusion that the two cases are legally the same? To answer the question, the judge needs to determine how strong the analogy is if showing that the two cases are legally the same is its purpose. We can illustrate the judge's situation by thinking about a reasoner who is confronted with an argument by analogy supporting a conclusion that she needs to either accept or reject based on the argument. Through the presentation of the conclusion, the argument shows the purpose for which its analogy was drawn. The reasoner must evaluate whether the analogy can fulfill this purpose.Footnote ⁸² Take this example:

The premise of an argument is meant to support its conclusion. Therefore, this analogy clearly has the purpose to support a copy and substitution that results in the claim that “the well-being of their wards is of extreme importance to the emotional well-being of kindergarten teachers.” To understand the argument (a necessary step for evaluating it), we need to see how the mapping between kindergarten teachers and mothers could support the substitution.

This purpose-constraint guides the mapping in two ways. First, it guides a mapping from the claim that is supposed to be supported about kindergarten teachers to its best correspondence in mothers. We perform the mapping and get “the well-being of their children is of extreme importance to the emotional well-being of mothers.”

Second, it guides attempted mappings between kindergarten teachers and mothers for all those aspects of mothers that we consider to be inferentially connected to the aspect that the well-being of their children is of extreme emotional importance to their own well-being. One sign for this might be that mothers show strong signs of stress and depression when their children are not well. The purpose-constraint guides us to try a mapping of this aspect onto kindergarten teachers. We might observe kindergarten teachers and realize that they have more fights with their partners at home when their wards are not well. The success of this mapping will be measured in how well the surface- and structural-constraints can be observed. The more successful mappings can be performed, the better the support for the “copy and substitute” that the argument suggests. These are similarities that are relevant for the argument. To understand the argument, we search for as many relevant successful mappings as possible to see how strong the support of the argument for its conclusion is. The conclusion of an argument guides the mapping by giving the analogy a purpose.Footnote ⁸³

In reasoning by precedent according to the analogy-approach, the conclusion of the analogical argument is simply that the two cases are similar in the legally relevant ways and that there is no relevant difference (that they are legally the same). How can this provide much guidance?Footnote ⁸⁴

To answer this question, I should point out that the analogy-approach has more shortcomings than just the problem of explaining precedent-constraint. Just like the straightforward rule-approach did not adequately reflect the practice of distinguishing, the original analogy-approach makes no reference to the important role that the precedent-opinion and the ratio play. But an adequate account of reasoning by precedent should take this role seriously for at least two reasons: first, not integrating opinions and rationes decidendi makes the account less than descriptively adequate. Second, the arguments presented in the precedent-opinion are likely to give rise to expectations about how later cases will be treated—expectations that go further than those that could be generated by the decision on its own. If the doctrine of precedent is supposed to realize rule-of-law values, then the method of reasoning by precedent should lead to decisions that usually do not disappoint these expectations, at least where they are reasonable.

How can opinions and rationes be integrated into an account of reasoning by precedent that is based on analogical mapping? I will develop the suggestion that they play a role as authoritative descriptions of the source-analog, that is, the precedent. Thereby, they also give the precedent its constraining force.

C. Precedent-Opinions as Authoritatively Described Source-Analogs

When she is reasoning according to the analogy-approach, the present judge finds herself in a position where she is trying to evaluate an argument by analogy, all parts of which—the source, target, and conclusion—have been given to her in the form of the precedent-case, the present-case, and the decision in the precedent. The present judge informs herself about the precedent-case by reading the opinion. It contains a detailed description of the precedent-case that gives special emphasis to those aspects that the precedent judge wanted to present as crucial for her decision. Here, the precedent judge presents the official reasons for her decision. That means that the precedent is not like most other source-analogs. For one thing, through the opinion's description, the precedent judge provides the present judge with an indication of how the precedent-case appeared from her point of view (or at least from the point of view she wanted to document). She rhetorically highlights those aspects that she thought important, for example by mentioning them often or by giving them an important place in one of her arguments. Sometimes, she also orders them into clear categories. At other times, the category they belong to will not be clear, as when the fact that it was the mother who witnessed the death of her child in Dillon v. Legg is vaguely specified as a close relationship.Footnote ⁸⁵ Other facts will be omitted or mentioned in passing and never be picked up again, receiving little attention, making it clear that the precedent judge did not think them important. Still others will be highlighted as the basis for possible but rejected counter-considerations.

However, that a precedent-opinion emphasizes some of the precedent-case's aspects is not all. There are many arguments by analogy that do this, because by emphasizing certain aspects in the source-analog it is possible to direct the audience's attention to them and raise the probability that they will attempt to map them. For example, imagine we had a mutual friend who was a kindergarten teacher, and I was simply trying to convince you that we should take seriously the emotional harm she might suffer when her ward gets hurt:

If we think about opinions, it turns out that they make for source-analog that are quite different from other kinds of source-analogs: they do not just describe the source-analog, their description is authoritative.Footnote ⁸⁶

First, a vigilant audience member presented with the sad-kindergarten-teacher argument might legitimately criticize the characterization of the source-analog. They might claim that some mothers are callous and care little about the well-being of their children. This would be a possible (though not plausible) objection to the argument. Not so with precedent-opinions. In the opinion, the precedent judge describes the facts of the case authoritatively; those facts make up the precedent-case, even if it turns out that the real-life situation of the case did not contain them after all.Footnote ⁸⁷ Also, if a judge writes that an aspect belongs into a certain category (whether this category is specific or not) then it is an aspect of the precedent-case that it has an aspect that falls under this category. Thus, the opinion authoritatively determines the aspects of the precedent-as-source-analog.

Second, an audience member of the sad-kindergarten-teacher argument might criticize it by denying the inferential connections that are implied by the description of the source-analog. For example, they might object that mothers usually get blamed when their children come to harm, and that is what stresses them and causes the fights. So even if kindergarten teachers show similar behaviors, that is no reason to think that their emotional well-being depends (directly) on the well-being of their wards—they simply do not like to be yelled at. This, too, is not possible with precedent-opinions. Opinions also authoritatively determine which of the precedent-case's aspects are legally relevant for the precedent-decision. They do so because they contain a description of the argumentative structure that, according to the precedent judge, presents the reasons based on which the decision was made.

Now, it is generally acknowledged that judges make bad decisions, and that these bad decisions still constrain. Levenbook has translated this into the claim that not every precedent-decision is justifiable, but even unjustifiable decisions are authoritative. Judges make mistakes. They might think that an aspect of a precedent-case presents a reason for a decision when it really does not. It might then be that the decision was unjustifiable when the judge made it. With this understanding of reasons, it is not plausible to say that a present-case has to be decided by following a precedent-case if the reasons that justified the precedent-decision also apply to it.Footnote ⁸⁸ But while it might be true that a specific precedent-opinion fails to present reasons that together justify the precedent-decision, it most often does present arguments (albeit possibly bad ones). Arguments purport to present reasons for a conclusion but may well fail to do so. These arguments cite some aspects rather than others as being the grounds of the decision. By writing the opinion, the judge marks these aspects as relevant for her decision. And because our judge operates in a common law system, her marking them as relevant is also authoritative. We can say that she has put a mark of authority on them so that from now on, the aspects that she presented as reasons for her decision are now legally relevant, relevant by judicial authority.Footnote ⁸⁹ The same goes for those aspects the judge describes as bases for arguments against her decision and that she considered outweighed or defeated by the arguments in its favor. As Lamond points out, considerations based on these aspects are now authoritatively determined as too weak to outweigh the arguments for the decision. There are therefore two critical questions that are helpful for evaluating normal arguments by analogy but excluded in precedents: the question whether the source-analog is correctly described and the question whether the properties that the description of the source-analog presents as relevant are actually relevant.Footnote ⁹⁰

These authoritative determinations significantly affect the purpose-constraint of the analogical argument in reasoning by precedent. I already pointed out that the present judge is evaluating an argument with the conclusion that the two cases are the same in the legally relevant ways. The purpose-constraint with respect to this argument therefore requires that the legally relevant aspects in the source-analog, i.e., the precedent-case, are mapped. And those are the ones that have been authoritatively determined as such in the precedent-opinion. So the purpose-constraint in reasoning by precedent guides attempted mappings from these aspects to the present-case. The success of these mappings is judged according to the other two constraints. If corresponding aspects can be found, the judge has found a similarity that is legally relevant. It is legally relevant because it contains, as one correspondent, an aspect that was authoritatively marked as legally relevant by the precedent judge. The question of which similarities are legally relevant can therefore be answered in the amended analogy-approach of reasoning by precedent.

Now, if the judge is honestly interested in determining whether the precedent gives her a reason to follow, she will try to pick out those correspondents that make for the best mappings because that will allow her to see the strongest case for following she can understand. Overall, the attempt to evaluate the analogical argument by precedent then leads the judge to perform the most successful mapping of precedent-case and present-case that she can accomplish.

The ratio plays an important role here, however well described it is. Those aspects that can be identified as belonging to the ratio, for example because they are presented as an integral part of the argument, are most important for the success of the mapping. For these aspects, the judge must be able to find a successful correspondence in the present-case according to the surface- and structural-constraints. The opinion's determinations are authoritative; therefore, the way it characterizes the precedent's aspects will determine whether these constraints can be met, even if these characterizations are vague. Take for example the close-relationship aspect in Dillon v. Legg. Here, the surface-constraint applies for the relation: both parts of the correspondence must fall into the category “close relationship.” The structural-constraints apply for the fillers of the relation throughout the cases.

If no mapping can be accomplished with respect to a part of the ratio then the argument for legal sameness fails: the cases are not similar in the legally relevant ways. Otherwise, the mapping was successful. Then the two cases are similar in a legally relevant way and the existence of the precedent is an authoritative reason for following. Of course, it is not necessarily a binding reason—distinguishing is possible, just as it was with the rule-approach.

Notice that the opinion does not need to provide a rule to guide the mapping in this way. Neither does the judge need to determine such a rule by filling in where information is missing in a vague opinion. Rather, the judge uses whatever information is provided as guidelines for her own mappings. She simply chooses those aspects of the present-case that to her seem to enable the most successful mappings for the aspects in the precedent-case. This is not meant to deny that analogical reasoning by precedent imports first-order reasoning. Where the opinion is vague, the judge's own first-order reasoning helps her determine whether she considers some mapping successful. For example, the specification “close relation” might lead her to recognize that the best mapping for the aspect “mother” available in the present-case is the aspect “kindergarten teacher.” She will have to decide whether she thinks that “kindergarten teacher” and “mother” are similar enough to count. For this she will have to consider all the ways in which the mapping succeeds (aspects referring to emotions can be mapped) and fails (aspects referring to biology cannot be mapped).

With respect to the vague parts of opinions, the analogy-approach imports first-order reasoning, just like the rule-approach. However, where the opinion does give a definite category, the analogy-approach excludes discretion, just like the rule-approach. This is so because when a definite category is provided, there is a successful mapping whenever an aspect in the present-case falls under this category simply because the surface-constraint is maximally satisfied. Here, the opinion has made clear that it is falling under this category that is relevant. Therefore, the opinion provides the judge with exactly as much or as little guidance when she reasons analogically as it does when she tries to determine a rule.

Remember also that a judge who is reasoning according to the rule-approach must make sure that a rule she chooses fits with the surrounding law. In the same vein, a judge cannot claim that a mapping is successful or unsuccessful if there are parts of the surrounding law that forbid treating the respective aspects in precedent and present-case as the same or as different (as for example an anti-discrimination law might).

By integrating opinions and rationes into the analogy-approach of reasoning by precedent in this way we have hit two birds with one stone: the analogy-approach is now descriptively more satisfying than it was before it was amended. Furthermore, the analogy-approach is now able to explain how precedents serve to constrain judicial decision-making. The authoritative determination of the precedent-case in the opinion is what constrains the judge. Unlike arguers, who can choose their description of the source-analog to influence whether the mapping will succeed, the present judge must take the description of her source-analog from the precedent judge. The opinion determines which aspects of the precedent-case must be mapped successfully for the whole mapping of the two cases to be successful.

To represent the importance of opinion and ratio, the argument-scheme for the analogy-approach of reasoning by precedent must be amended in the following way:

A. The First Difference and Advantage: Possible Futures

Arguments by analogy are used not only because of their justificatory power, but also because of their rhetorical strength. Analogies are rhetorically effective because they encourage those who are trying to understand them to restructure their understanding of the analogs in a way that makes them appear more similar.Footnote ⁹³ The mere attempt to understand an analogy lends some plausibility to the analogy in the eyes of the reasoner. Why this is so can most easily be explained by using the rhetorical concept of presence.Footnote ⁹⁴

There are many aspects to any subject matter. For example, there are many things that one can say about kindergarten teachers. Nonetheless, if I ask you to think of kindergarten teachers, there will be some things that come to your mind more readily than others. You might first think of the fact that kindergarten teachers take care of children. Whatever comes into your head first, that is what has most presence in your mind. The aspects of kindergarten teachers that have the most presence in your mind will greatly influence your impression of what a kind of person a kindergarten teacher is—a loving caretaker, a professional, a chronically underpaid and likely exploited worker. It is easier for you to reason with those aspects of a subject matter that have a lot of presence in your mind than with aspects that, for you, float in the background. Therefore, it is easier for you to understand and accept an argument whose premises play upon those aspects that are already present to you. When you perform an analogical mapping to understand the sad-kindergarten-teacher argument, this (temporarily) influences the presence that certain aspects of mothers and kindergarten teachers have in your mind: you pay attention to those aspects that you are guided to map according to the purpose-constraint. As a result, you come to think of kindergarten teachers and mothers differently. The ways in which they are relevantly similar for the argument step into the foreground. The argument by analogy has led you to construct kindergarten teachers as being similar to mothers with respect to their emotional reactions to the well-being of the young people in their care. You concentrate on those reactions rather than, for example, the fact that kindergarten teachers are paid to do their job. Kindergarten teachers and mothers now have a similarity you might have never paid attention to before. That is why arguments by analogy are such persuasive rhetorical tools. They make you want to understand them and understanding them makes you construct an understanding of the target and source as similar that also incorporates the conclusion as plausible.Footnote ⁹⁵ When you evaluate such an argument, you also evaluate the adequacy of the understanding of the analogs that this argument encourages you to adopt. The trick is that this understanding has been constructed by you under the purpose-constraint, and therefore it is an understanding that is plausible to you.

There are some important differences between normal reasoning by analogy and analogical reasoning by precedent. What stays the same, however, is the rhetorical effect that the attempt to map the two cases together has on the judge's understanding of them. In analogical reasoning by precedent, the present judge can use this effect to construct and evaluate an understanding of the present-case that makes it as plausible as possible that it is similar to the precedent-case in a legally relevant way. This allows her to see the most persuasive version of how the present-case must be understood if she decided to follow. By trying to understand how the two cases could be mapped together, she constructs the view of the present-case that is most likely to convince her that the present-case is like the precedent-case, simply because this view is the outcome of her attempt to fulfill the purpose-constraint and therefore gives presence to the successful mappings between the cases.

Interestingly, this process of considering how the two cases could be similar also shows her how the precedent-case must be understood if the mapping were to be considered successful. The judge is constrained in her understanding of the precedent-case by the way the opinion describes it. But where the opinion is vague, her attempt to map the two cases together will lead her to construct an understanding of the precedent-case so that it fits the present-case. Take a case in which the best correspondence for the “close relationship” in Dillon v. Legg is the relationship between a kindergarten teacher and her ward. Then the mapping can be considered successful only if “close relationship” is interpreted widely, and not limited to blood relations. So, if the judge says that Dillon v. Legg needs to be followed in the present-case, then she needs to commit to reading “close relationship” as a category at least that broad. The vaguer the precedent-opinion, the stronger this effect will be because it will apply to many aspects of the precedent-case. By performing an analogical mapping between the two cases, the judge comes to see both the precedent-case and the present-case in the way they must be understood if she decided to follow. She also sees which understanding of both cases she must reject if she decides not to follow the past case. If this understanding is importantly different from how she understood either case at first sight, these differences will attract her attention. If she recognizes that following in a present-case would necessitate a change in the way she understands the precedent, then she also recognizes that her decision either way will influence what that precedent will stand for in the future. The mapping of precedent-case and present-case might be consistent with the precedent-opinion and the surrounding law. She now sees that, for example, the viewpoint of the party who brought up the precedent is not absurd, that there is a way to see the precedent as applicable. Then the fact that it does justice to the present-case will weigh in as an argument for changing her understanding of the precedent. She might conclude that the two cases are relevantly similar, even though the present-case is not relevantly similar to the precedent-case as the judge originally understood it. Without the mapping, she might not have become aware of this.

Analogical mapping brings to the attention of the judge ways to understand the precedent that she might not have thought of had she not tried to achieve the best possible mapping with the present-case. The rhetorical force of analogies therefore generates an advantage, allowing her to make a more balanced decision based on more considerations. This benefit might not be available if the judge reasons according to the rule-approach. Here she chooses a rule by considering the precedent alone and how she believes it can best be understood as good law. She might not even think of other ways to understand the precedent that would lead to a different decision. She concentrates on the precedent-case and the normative reasons that she thinks are relevant for determining the rule connected to it. Only when the rule has been chosen does the present-case enter the picture, and by then the question how the precedent should be understood is closed. The analogy-approach draws the judge's attention to how her decision to follow or to reject following might influence the development of the common law.

B. The Second Difference and Advantage: Distinguishing and Relevant Differences

So far, I have not spent much time discussing the reasoning involved in deciding whether to distinguish. I mentioned that a completely pure rule-approach would require the judge to make a list of categories containing all the properties of the precedent-case that are the basis for excluded reasons against the precedent-decision. The present-case could then be distinguished only based on reasons that cite properties that are not on that list. This reading of the rule-approach would be consistent with what Lamond says about distinguishing, namely, that any reason based on a fact already present in the precedent-case is excluded from consideration when it comes to distinguishing. However, it should be easy to see that the analogy-approach can also accommodate this because it only allows distinguishing based on relevant differences between precedent and present-case.

All this is about what cannot be a reason for distinguishing. But what is a reason for distinguishing? This is an important question because the injustice done by wrongly following a past case is just as great as the injustice done by wrongly distinguishing a present-case. As Rigoni points out, there are always some differences between two cases, but these differences can only warrant distinguishing if the difference-constituting fact provides a reason against the decision that would result from following. Footnote ⁹⁶ A whole paper could be written on the question when the decision to distinguish could be considered justified in the analogy-approach, but space permits me only to say very little. What might be obvious is that when the judge first hears the present-case, it will have certain aspects that she notices as pertinent or that the lawyer of one of the parties points out as pertinent. Any one of these aspects that is not included in the mapping of the two cases warrants special attention. If it is a basis for reasons against following the precedent-decision, then it is a possible ground for distinguishing. This is why it would make little sense for a judge to analogically map only those aspects of the precedent-case onto the present-case that the opinion clearly marks as part of the ratio. All aspects of the present-case that can be successfully included in the mapping are similarities between present-case and precedent-case that are relevant at least insofar as all the successfully mapped aspects cannot be the basis for relevant differences.

Further, analogical mapping can bring attention to relevant differences that might not be initially obvious. Cognitive scientists have found that it is easier to see differences between analogs that have an overall common structure than between ones whose structures had not been mapped.Footnote ⁹⁷ This makes sense from a rhetorical point of view, because the process of mapping makes the mapped aspects more present to the reasoner, making it easier to include them in further reasoning. By influencing how present aspects are, the process of analogical mapping leads not only to the recognition of similarities, but also to that of differences. In one sense, this was already implied by what I said earlier: the purpose-constraint might lead to the attempted mapping of a fact that is clearly part of the ratio and this mapping might fail. However, there is a further way in which mapping may help with the recognition of differences—this time relevant differences that lead to distinguishing. Take again the example of MacDonald v. Ortho Pharmaceutical Corp. Footnote ⁹⁸ The reader will remember that the opinion gave a clear rule-like formulation: the duty to warn the consumer does not apply if a competent intermediary has been informed of the risk. The court cited the case of McEwen v. Ortho Pharmaceutical Corp., according to which both prescribing and treating physicians count as such intermediaries.Footnote ⁹⁹ This could have been the end of the decision.

However, according to the analogy-approach, even a rule-like formulation is used as guidance for a mapping, instead of just for an application of the categories. And a more detailed mapping between the physician in MacDonald and the physician in McEwen reveals the relevant difference between the two cases. In McEwen, the court treated both prescribing and treating physicians as competent intermediaries. As one reason, it mentioned that the treating doctor could see how the patient would react to medication over the long term and determine whether the patient should stop taking the medication. The court decided to distinguish in MacDonald. It did so because the patient decides relatively autonomously when to take birth control. It is plausible that the attempt to map a birth-control patient-doctor relationship onto the relationship between patient and doctor as described in McEwen could have made this difference apparent.

The mapping was successful when it came to the category that the court had made out (there must be a physician who is warned), so there was a relevant similarity simply because the surface-constraint was maximally satisfied. But the mapping also gave presence to the role of the physicians and thereby to the difference between these aspects in the two cases. Of course the fact that a user of birth-control pills is more autonomous than users of other prescription drugs could have been noticed by a judge who was reasoning according to the rule-approach. I do not mean to claim that there are grounds for distinguishing that can only be noticed through analogical mapping. I claim is that analogical mapping is more promising when it comes to drawing attention to differences that lie beneath relevant similarities and can justify distinguishing. The success of a mapping is judged by the degree to which it fulfills the structural-constraints as well as the surface-constraint. Determining whether the structural-constraints can be fulfilled requires the judge to go into detail to see whether the relations of the mapped attribute can also be mapped, and their fillers, and so on. Such detail is not required in the rule-approach. According to the rule-approach, the judge constructs a category and determines whether this category can be applied to the present-case. If it can, then that is a success.