Natural reasoning and deductive reasoning compared

If all judges were perfect reasoners, the natural model of judicial decision-making clearly would be superior to the rule model. Under the natural model, judges attempt to identify the best outcome for each individual case.

We have already described in Chapter 1 some of the errors associated with rule-governed decision-making. Rules are distinguished by their determinacy, which enables the various decision-makers who apply the rules to understand readily what the rule requires. Yet, rules that are determinate enough to govern decision-making effectively will, inevitably, dictate a certain number of results that are inconsistent with the rules' underlying purposes. A rule stating that bears kept as pets in residential neighborhoods are nuisances per se is almost cer­tainly a good rule, but it may sometimes result in the exclusions of a completely harmless bear.

Another problem associated with general, determinate rules is the problem of bad rules. Determinate rules prevent certain types of error by translating the rule-maker's expertise into definite instruc­tions about what to do or decide, reducing the deliberative costs of decision-making, and enabling actors to coordinate their conduct around easily comprehensible prescriptions for action. They also cause error by prescribing wrong outcomes in some cases, as just described. They are justified when, judged by the values on which they are based, they prevent more errors than they cause. In some cases, however, the applicable rule may fail this test, due either to faulty rule-making or to changed circumstances over time.

Common law rules crafted by judges are particularly likely to lack jus­tification from the outset. The immediate task of a judge is to settle a single dispute.

Despite the various difficulties of a rule model of rule-based decision­making by judges, we believe it is justified in a significant range of cases. The comparison is not between rule-following and ideal reasoning, but between the shortcomings of rule-following and the shortcomings of unconstrained reasoning that does not yield reliable rules for future actors and future decision-makers. Deductive reasoning from rules is preferable for the range of situations in which there is reason to think that the sum of errors by judges and actors will be lower under a regime of rules than under a regime of all-things-considered judicial decision-making.

Examining this comparison more closely, the capacity of determinate rules to prevent errors depends primarily on the expertise of rule­makers and the coordination value that rules offer to actors who rely on them to guide their decisions. Looking first at rule-makers, there is no reason to assume that an earlier judge considering whether a mor­tuary, or a pet bear, in a residential neighborhood is a nuisance, has greater expertise than a later judge considering a similar case. In fact, the later judge may have a better sense of prevailing shared expecta­tions about land use at the time her case arises. Focusing only on these two judges, deciding their two cases, comparative expertise does not support treating a rule set forth by the first judge as authoritative. On the other hand, an earlier decision announcing a rule may gain epis- temic weight over time: if multiple judges endorse the first judge's rule, there may be reason to think that the rule reflects judicial expertise.

A further consideration is that judicial rules are not addressed only to future judges. In a legal system in which legal decisions and the reasons judges give for their conclusions are well publicized and widely consulted, the relevant comparison in thinking about judges as rule­makers is between the expertise of judges and the expertise of private actors deciding what to do in various situations. The judge has two epistemic advantages in this situation: she has heard evidence and arguments from opposing parties, and she has no personal stake in the outcome. Private actors, in contrast, may be less well informed, and their judgment may be affected by self-interest. On the other hand, in some settings private actors may have a better understanding of the nature of their own activities than even a well-briefed judge. If so, future actors may be better off innovating for themselves than follow­ing judicial rules.

A stronger argument in favor of treating judicial rules as authorita­tive is based on the settlement and coordination benefits that follow from the existence of rules that are treated as binding by judges and other decision-makers. If prior courts have developed a per se rule that mortuaries in residential areas should not be enjoined as nuisances if they pass municipal safety inspections, and later courts have accepted the rule as authoritative, mortuary operators know what to do to avoid a nuisance claim, and potential buyers know that they may end up with mortuaries as neighbors. All parties can adjust their plans accordingly. Future judicial decision-making is also simplified, because the question whether a mortuary can do business in a residential neighborhood is settled.

Notice that a judicial standard does not have the same effect as a rule. Current judges may accept a standard as authoritative, but its author­ity, and its ability to guide future actors and litigants, is limited to the matters it settles in a determinate way.

The rule model of common law decision-making recognizes the power of judges to announce authoritative rules and the duty of later judges either to follow precedent rules or to set them aside in their entirety. Not all judicial decisions will qualify as rules: some decisions will be too narrow and case-specific to provide useful settlement, and some rulings will be phrased as standards that guide but do not preempt natural reasoning. When judges do make rules, not all of their rules will be good ones: judges lack the fact-finding capacity of legislatures, and they may be distracted by special features of the cases that come before them. Nevertheless, the rule model generates an additional set of rule-making authorities, capable of increasing the breadth of settlement and the level of coordination within a community.

We have already noted that judges who decide cases based on natu­ral reasoning alone may give weight to prior judicial statements, on the ground that a practice of honoring prior statements will enable actors to plan more effectively and coordinate their actions with the actions of others.

One difficulty with this outcome is that the current judge may be wrong: her assessment may have been affected by the artful case presented by the mortuary and the failure of the objecting homeowners to focus on the coordination benefits of consistent decision-making over time. A further complication is that psychological research suggests that judges who take a natural approach to rules will tend to undervalue the long-term values associated with rule-following, such as coordination of private action, in comparison to more immediate concerns about the effect of the rule on current litigants. The difficulty is cognitive: decision-makers tend to focus on facts that are especially salient, and in doing so, to disregard or undervalue background probabilities, such as the likelihood of future coordination around rules. This bias, which Kahneman and Tversky have called the ‘availability heuristic', is one of a number of cognitive biases that facilitate human decision-making but also distort reasoning in systematic ways (see Tversky and Kahneman 1982; Schwarz and Vaughn 2002). As a consequence of this bias, tangi­ble facts are likely to move quickly to the forefront of decision-makers' minds, while statistical regularities remain unnoted.

This type of bias in favor of salient facts suggests that, although rule­sensitive particularist judges may assign value to the coordination ben­efits of consistent decision-making in the course of natural reasoning, the value they assign is likely to decline over time. Assume that judges will sometimes make mistakes, and that they will err most often by giving more weight to the readily accessible facts of the cases before them than they give to less salient background considerations such as the settlement and coordination value associated with consistent application of rules stated in past opinions. Under these very plausible assumptions, the value of consistent application of rules will in fact decrease, as judges reduce their predictions of the extent to which other rule-sensitive particularist judges will follow the rules.

Suppose, for example, that a judge has before her a case in which homeowners seek to enjoin a mortuary from opening in their neigh­borhood. Her initial conclusion, reached through a process of reflective equilibrium, is that no plausible moral principle supports the exclusion of mortuaries from residential neighborhoods. She then finds a case in which a prior judge in the same jurisdiction announced a rule that mortuaries in residential neighborhoods are nuisances per se. As a rule-sensitive particularism she may, despite her initial conclusion, give weight to the possibility that homeowners have relied on the prior case in purchasing property and to the coordination benefits of having a reliable rule. Yet, if she believes that most other judges are also rule­sensitive particularists, she will assume that over time, as judges vary in the weight they assign to these rules in their reasoning, rule values will lose importance. Telescoping forward, she may simply decide to exclude rule values from consideration. In this way, rule-sensitive particularism becomes nearly identical in effect to pure particularism, which takes no account of the value of rules.

In a mixed population of judges, some inclined to treat precedent rules as authoritative and others acting as rule-sensitive particularists, the coordinating value of rules will still be unstable. The potential harm that particularistic decisions will have on the coordination value of rules may cause rule-followers to reassess the value of their own obedi­ence, and to turn toward particularism in difficult cases. In theory, rule-sensitive particularist judges should take this possibility into account as a reason to increase the value they place on coordination through regular enforcement of rules. Yet, for the psychological rea­sons just discussed, they may tend to discount this type of extenuated consideration when assessing what seem to be compelling facts of a particular case.

In contrast, the rule model of judicial decision-making eliminates the type of cognitive error that can lead to a decline in the coordina­tion value of judicial decisions. The rule model preempts all-things- considered reasoning within the boundaries of the applicable rule. In doing so, it both captures the potential coordination value of the rule and protects against judicial erosion of that value over time.

A further advantage of the rule model is that it reduces the complex­ity of the decision-making process. A judge following the rule model must interpret rules announced in the past to determine whether they apply to the case before her, but once she has decided that a certain rule is applicable, all that remains is to follow the rule's prescrip­tion. For a judge engaged in natural reasoning, the existence of a precedent rule complicates rather than simplifies decision-making. A rule-sensitive particularist judge faced with a question that may have been adjudicated in the past must engage in interpretation of previously announced rules; then if any apply, she must weigh the settlement and coordination benefits the rule might provide against the costs of errors she believes would follow from applying the rule to the case before her.

We believe that, in a certain range of cases, the rule model of judicial decision-making offers significant benefits in comparison with natural reasoning. For a number of reasons, however, rule-based decision­making will never be, and should not be, the exclusive method of the common law. The primary task of a judge is to resolve a particular dispute, and in some situations the judge may rightly believe that the case before her is not sufficiently representative of conduct likely to result in disputes to provide the basis for a rule. In other situations, she may rightly believe that a rule is not the best method for governing the type of conduct in question. The benefits of settlement and coordina­tion may be comparatively low, and the current judge may prefer to leave future judges free to take account of factual variations or chang­ing norms. Yet there are also settings in which regular application of a determinate rule announced in prior cases will reduce the sum of moral and practical errors in judicial decision-making and in private action that occurs within the framework of law.

In any event, the rule model of judicial decision-making does not pro­vide a full description of what judges do. Judges can and must engage in natural reasoning in the interstices of rules. In addition, they may decline, with reason, to exercise the full range of legislative power that the rule model makes available to them. As we will describe at length in Chapter 6, they also engage in a variety of methods and practices that do not fit either the natural model or the rule model of decision-making.

The main point we wish to make in this chapter is not that one or the other of the two forms of reasoning we have described is or should be predominant in judicial practice, but that the models we have described - the natural model and the rule model - are the only plausi­ble models of judicial decision-making. Judges also make use of stand­ards, which guide reasoning without dictating outcomes, but standards are simply hybrid combinations of natural and rule-based reasoning. A further point is that neither the natural model of reasoning nor the rule model of reasoning is a specially ‘legal' form of reasoning. Both rely on reasoning techniques used by decision-makers in all contexts: natural reasoning, empirical reasoning, and, in the case of the rule model, deduction from authoritative rules. These are the only tools that judges need and the only tools they in fact use.

5.4