In 1966 the U.S. Supreme Court wrote, “The basic purpose of a trial is the determination of truth.”Footnote ¹ This is Larry Laudan's guiding premise in his “essay on legal epistemology.” Without ascertaining the facts about a crime, he writes, it is impossible to achieve justice, since a just resolution crucially depends on correctly figuring out who did what to whom.Footnote ² Thus, he continues, “it is entirely fitting to ask whether the procedures and rules that govern a trial are genuinely truth-conducive.”Footnote ³ In chapter 1 of the book, Laudan identifies one of the most important and legitimate methods for finding truth, namely, ensuring that the jury hears all and only relevant evidence. Laudan bemoans the fact, however, that “legal texts and the practices of courts routinely flout” this principle.Footnote ⁴ Much of the rest of the book is devoted to the other tests for admissibility that the system imposes, tests that Laudan often regards as misguided.

In addition to the value of truth acquisition (or error reduction) in a trial system, Laudan identifies two other values: first, error distribution, which seeks to balance the relative frequency of false acquittals (of the guilty) to false convictions (of the innocent); and second, nonepistemic policy values, which vindicate such ends as efficient use of resources and protecting the rights of defendants. Laudan argues that our criminal justice system takes a flawed approach to these projects by focusing too much on ensuring that innocent defendants are not wrongly convicted and protecting the so-called “rights” of criminal suspects.

In chapters 2 to 4, Laudan addresses what he terms the “error distribution doctrine”: a set of concepts and precepts, such as the beyond-a-reasonable-doubt (BARD) standard of proof and the presumption of innocence, that are explicitly “designed to safeguard the fate of the defendant by making it hard to convict any but the most obviously guilty.”Footnote ⁵ Laudan argues that their application in American law achieves undesirable epistemic results.Footnote ⁶

Laudan argues that the BARD standard is best seen as a device that reflects a “social contract” concerning the acceptable ratio of true acquittals to false convictions.Footnote ⁷ Because the standard of proof is (or should be) at heart a method for distributing errors, Laudan is greatly concerned about the ambiguity embodied in the BARD standard. He reviews conflicting jurisprudence in which different courts define that standard in various unsatisfactory ways or even refuse to define it at all.Footnote ⁸ Laudan champions reform in the standard of proof so that it will focus on the objective evidence rather than jurors' subjective beliefs and doubts. He specifically wants an interpretation of BARD to be presented to jurors that will reflect the social contract concerning the acceptable ratio of true acquittals to false convictions.

In the second part of Truth, Error, and Criminal Law, called “Flawed Rules of Evidence and Procedure,” Laudan promotes the repeal or reform of numerous judicial rules and procedures that he believes serve as major obstacles to truth determination. The rules include some that most practicing lawyers would consider basic tenets of our criminal justice system, such as the exclusion of evidence obtained in violation of the defendant's rights, the inadmissibility of some hearsay evidence, and the prohibition on drawing inferences from a defendant's failure to testify at trial.Footnote ⁹ Thus Laudan's proposals would overhaul the basic rules by which criminal trials are conducted in the name of achieving better epistemic results.Footnote ¹⁰

Laudan contends that once society selects a standard of proof that reflects an acceptable ratio of false acquittals to false convictions and once a judge appropriately instructs the jury about that standard, rules that further impede the prosecution's efforts to obtain a conviction are improper and pointlessly hamper the search for truth.Footnote ¹¹ This argument reflects Laudan's fundamental preference for a “unitary approach,” as we call it, to the distribution of errors, under which distributionist goals are accomplished only by the court's instruction to the jury about the standard of proof. Laudan rejects what might be termed a “diffuse” approach to error distribution pursuant to which society's distributionist goals are accomplished by the standard-of-proof instruction along with several other procedures and rules, some of which could be targeted to particular problems with jurors' reasoning.

The core of Laudan's opposition to a diffuse approach is his argument that evidentiary rules that benefit defendants do so to an unknown degree. Because we have little empirical evidence about, for instance, jurors' estimation of the probative value of coerced confessions, Laudan claims that we cannot know the effect of a rule excluding evidence of such confessions from trial.Footnote ¹² In his account, the various exclusionary rules with distributionist effects thereby end up muddying the standard of proof and unfairly benefiting the defendant:

The problem . . . is that adding additional dollops of the [benefit of doubt] by adopting defendant-friendly rules may produce a [benefit of doubt] that is massively greater than the acceptable one. . . . It follows that, if we discover that the current [standard of proof] confers less [benefit of doubt] than society regards as desirable, the only appropriate remedy is to raise the existing standard of proof, not to skew the rules of evidence in favor of the defendant.Footnote ¹³

A bedrock principle in Laudan's analysis is his somewhat apostate view that defendants have a huge advantage in the litigation of criminal cases.Footnote ¹⁴ He disputes the notion that prodefendant rules are required to “level the playing field” between well-funded prosecutors and individual defendants.Footnote ¹⁵ Rather, he asserts, “[t]he playing field . . . already begins tilted heavily toward the defendant by virtue” of the “enormous advantage” embodied in the standard of proof.Footnote ¹⁶ Indeed, Laudan avers that “[t]he actual handicapping in a criminal trial is precisely the opposite of what the distributionists would have us believe. From the outset of a criminal proceeding, the average guilty defendant is assured that, though guilty, he is unlikely to be convicted of a crime.”Footnote ¹⁷

One representative argument concerns a defendant's Fifth Amendment right to avoid incriminating himself and the so-called “Griffin rule,” which forbids the prosecution from commenting on the defendant's silence and requires the court to instruct the jurors not to draw any adverse inference therefrom.Footnote ¹⁸ Laudan suggests that the right to silence itself is an epistemically flawed rule: forcing defendants to take the stand would likely produce relevant evidence against guilty defendants, and the protections afforded to innocent defendants by the right to silence are likely to be slight.Footnote ¹⁹ He also attacks the Griffin rule, arguing that it is entirely reasonable to draw an adverse inference from a defendant's silence.Footnote ²⁰ Laudan addresses the possibility that jurors might attach too much significance to a defendant's silence by arguing that there is no empirical evidence of this problem. “Through a couple of centuries of American law” before Griffin, “juries showed themselves perfectly capable of seeing that guilt was often not proven even when the defendant said nothing in his own behalf.”Footnote ²¹ Thus, according to Laudan, the Griffin rule is faulty because we know it increases the likelihood of acquitting a guilty defendant, whereas we can only speculate that false convictions would occur in its absence.Footnote ²² Relying on his previous argument that evidentiary rules other than the standard of proof ought not to serve distributionist ends, he concludes that the Griffin rule should be abandoned.Footnote ²³

Laudan next addresses the constellation of rules that call for the exclusion from trial of illegally obtained evidence.Footnote ²⁴ The exclusionary rules make inadmissible confessions obtained by coercion or in violation of the Miranda rule, tangible evidence obtained in violation of a defendant's Fourth Amendment rights, and the fruits of evidence that falls into the preceding categories. Laudan argues that “[i]t is perfectly clear that the exclusion of germane, inculpatory evidence, on account of the way in which it was obtained, significantly increases the likelihood of false acquittals.”Footnote ²⁵ He does not carefully address the converse question—whether the admission of such evidence might lead to false convictions—except to assert baldly that an innocent person is highly unlikely to confess to a crime.Footnote ²⁶

The final chapter of Laudan's book addresses miscellaneous nonepistemic values that are used to justify evidentiary or procedural rules even though such rules may partially thwart the judicial system's truth-seeking function.Footnote ²⁷ Laudan somewhat mysteriously places into this category “[w]orries about the jury's inability to interpret evidence appropriately,” although he acknowledges that such worries are in fact epistemological. He concedes that commentators and courts invoke such concerns to justify a wide range of rules but he rejects those concerns as insufficient. He observes that almost no empirical studies test the implicit hypotheses about juror psychology that undergird these justifications and asserts that “[u]ntil such studies are available, our default assumption should be that relevant evidence is admissible.”Footnote ²⁸

We turn now to some critical reflections on Laudan's approach. We begin with Laudan's pervasive insistence on a unitary evidential approach under which the legal system's distributionist goals would be accomplished entirely and solely by the court's instruction about the standard of proof. It is this principle that causes Laudan to reject so many evidentiary rules and procedures: when those procedures have distributionist goals (or even effects), Laudan finds them inappropriate because that project, in his view, should be accomplished only by the standard-of-proof instruction. That argument has holes well worth exploring.

Initially, it is not clear why Laudan's unitary approach would be more successful than a diffuse one at achieving his much-emphasized goal of a certain and agreed-upon distribution of errors. As we discuss above, the core of Laudan's opposition to a diffuse approach is his argument that evidentiary rules that benefit defendants do so to an unknown degree. This is because, according to Laudan, we have little empirical evidence about, for instance, jurors' estimation of the probative value of coerced confessions. We cannot know the effect of a rule excluding evidence of such confessions from trial. But there is no clear reason why the effects of such rules are more or less certain than the effect of an instruction concerning the standard of proof. Laudan himself admits that he cannot fashion a BARD instruction he finds acceptable.

On the other hand, a diffuse approach to evidential matters has an important benefit: it permits the legal system specifically to target problematic types of evidence. As even Laudan acknowledges, if we know or suspect that jurors have trouble properly estimating the objective relevance of certain types of evidence (coerced confessions, for example), a diffuse system could deal with those problems by designing an evidentiary rule to govern the situation.

Relatedly, there is an important weakness in Laudan's contention that defendants currently enjoy a distributionist advantage well beyond that intended by the social contract. He assumes that the standard of proof applied by juries begins at approximately 90 percent (or wherever the formulations of BARD attempt to set it) and that the many evidentiary rules improperly push the true standard well above that level. But Laudan's opponents who argue that various evidentiary rules “level the playing field” for defendants may have a stronger point than Laudan concedes. These opponents do not literally mean that the rules should set the standard of proof at 50 percent—nobody seeks that result. Rather, they may fear—or have experience to suggest—that jurors do not really apply a high standard of proof to the prosecution's case at trial, regardless of the court's instructions. The first author of this review once interviewed after a trial members of a jury who admitted that they found the defendant guilty because the prosecution had presented nine pieces of evidence militating in favor of conviction while the defense presented only six pieces of exculpatory evidence. Even if society wants—and the court instructs—jurors to convict a defendant only if the evidence makes the jurors 90 percent certain of her guilt, jurors may fail to apply such a high standard because their certainty of guilt derives from other sources (various biases, the presumption that an innocent person is unlikely to be on trial, etc.). If it is true that juries in effect set a lower standard of proof than they are instructed to apply, we might want other rules to push the standard up toward our ideal (thus “leveling the playing field”).Footnote ³⁰

Turning to a second point, Laudan seeks a “social contract” about the proper level of the standard of proof. But he ignores a social contract that is already in place—the Constitution, at least as interpreted by the Supreme Court. It provides for targeted exclusionary rules, arguably because we as a people want certain types of evidence or methods of introducing evidence to be off limits. One example of this is the Sixth Amendment's confrontation clause. The clause provides that a criminal defendant “shall enjoy the right . . . to be confronted with the witnesses against him.”Footnote ³⁰ Recent Supreme Court jurisprudence holds that the confrontation clause establishes a per se bar on the admission of many out-of-court statements unless the declarant testifies.Footnote ³¹ There is no doubt that this constitutional rule sometimes prevents the introduction of relevant evidence, but that is the way we apparently want it. What has a better claim to being part of our social contract than something that appears in the Constitution? This is not to say that someone like Laudan should not advocate changes to the constitutional system. But since he pays homage to the notion of a social contract, it is worth observing that our social contract as presently constructed expressly embraces targeted exclusionary rules.Footnote ³²

A third issue we wish to raise concerns Laudan's view that the only appropriate criterion of evidence admissibility is the relevance test. Adding exclusionary rules beyond the relevance test, he maintains, is a practice not well designed to achieve epistemic ends. He defends this view, it will be recalled, on epistemological grounds, arguing that there is no better way to conduct the truth-determination business than by admitting all and only relevant evidence. We question this on the same grounds that Laudan endorses it, namely, epistemological grounds.

Is the procedure of admitting all and only relevant evidence the best procedure for getting at the truth—that is, for getting the jury to deliver true verdicts (verdicts that comport with the facts)? To examine this question, we must first ask what is meant by “relevance.” Laudan's answer is a familiar one: “[A] piece of information is relevant to the evaluation of a hypothesis just in case, if credible, it makes that hypothesis more or less probable than it was before.”Footnote ³³ The next question is: Granting the credibility of a piece of information, does relevance per se make its inclusion in the total evidence presented to a jury increase the likelihood that a correct conclusion will be reached? Inverting the formulation, does relevance per se make the exclusion of an item of evidence decrease the likelihood of obtaining a correct conclusion? Laudan answers this straightforwardly: “Excluding relevant but nonredundant evidence, for whatever reasons, decreases the likelihood that rational decision makers will reach a correct conclusion.”Footnote ³⁴ This is a thesis to which we take exception on purely theoretical grounds (the same sorts of grounds, as far as we can see, on which Laudan endorses it).

Another way of defining relevance, related to the one Laudan offers, is in terms of the ratio of likelihoods, or conditional probabilities, pertaining to an item of evidence E and the hypothesized event(s) to which it might be addressed. Consider two conditional probabilities (likelihoods): the probability that evidence E would have occurred conditional on target event X occurring; and the probability that E would have occurred conditional on X not occurring. If the two conditional probabilities, or likelihoods, are equal, then the likelihood ratio is 1.0, and evidence E is irrelevant. It is irrelevant because if one follows normative principles of probabilistic reasoning (Bayesian principles), a likelihood ratio of 1.0 will not lead a reasoner to revise her prior (subjective) probability for X. Thus there is no reason to admit irrelevant evidence at trial, because its (legitimate) effect would be nil. Items of relevant evidence should arguably be admitted at trial, however, because (assuming that the reasoner's prior subjective probabilities are neither 0 nor 1.0) it would lead to some sort of revision.

Now, the occurrence of some sort of revision in one's subjective estimate of X's occurrence does not have any obvious implication as to whether the revision in question is toward the truth or away from the truth. If X actually occurred, then increasing one's subjective probability in the occurrence of X is toward the truth; decreasing one's subjective probability in X is away from the truth. But it does not follow from revision per se in which direction the reasoner is carried: closer to or further from the truth. Laudan seems to assume that each item of relevant evidence (assuming its credibility, as Laudan puts it, or truth, as we put it) can be expected to lead the jury in the direction of truth. What justifies this assumption? Nothing, in our view.

To make the point clearly, we first need to distinguish systematically between subjective and objective probabilities. Subjective probabilities (including subjective conditional probabilities) are the decision-maker's own personal estimates, however well connected or poorly connected with the truth they may be. Objective probabilities are presumably facts of some sort pertaining to the world. There is little agreement about what sorts of facts these might be, especially when it comes to singular events (e.g., such-and-such a crime was committed at such-and-such a place). They might be facts concerning objective “chances.” Defining the notion of objective probability is a problem for anyone seeking to interpret the notion of relevance in general and its use by the Federal Rules of Evidence (FRE); it is not a problem peculiar to Laudan. But when FRE says that relevant evidence is “evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence,”Footnote ³⁵ what else can it be referring to other than objective probabilities? If it means, instead, to refer to somebody's subjective probabilities, whose subjective probabilities would these be? The judge's? The judge may have to decide on admissibility based on her own subjective probabilities, but the relevance rule is presumably intended to advise the judge what the objective standard for relevance is, and that must be something outside the judge's own head.

We assume, then, that the “objective” relevance or irrelevance of a piece of evidence E (with respect to a specified event X) is a matter of fact concerning the ratio of two objective likelihoods. The question is whether admitting all items of evidence E satisfying this condition is a maximally truth-conducive procedure with no other provisos added. This is what we mean to deny.

A theorem pertinent to this question is presented by Alvin Goldman and Moshe Shaked.Footnote ³⁶ They prove that if a Bayesian reasoner conditionalizes on any new piece of (true) evidence, starting from any prior that is neither 0 nor 1, and uses subjective likelihoods that match the objective likelihoods, then an objectively expected increase in truth possession is associated with her new posteriors. In other words, as a matter of mathematical necessity, there is always an objectively expected improvement in the direction of truth. This initially sounds like something that should please Laudan greatly. Admittedly, the theorem does not guarantee that new evidence always moves a rational (Bayesian) in the truth direction, only that movement toward the truth is objectively expected (probabilistically). However, Laudan should be content with this more modest finding. It seems to be a merely probabilistically qualified version of his thesis. However, careful note must be taken of the proviso that appears in the antecedent of this theorem, namely, that the agent's subjective likelihoods match the objective likelihoods. This additional condition is a very tough condition to meet. And it is a qualifying condition that goes beyond anything Laudan himself inserts into his general claim (at the theoretical level).Footnote ³⁷

As can readily be determined by calculation, decreases in truth possession can certainly be expected even when one conditionalizes on true evidence. This will occur when the subjective likelihood ratio goes in the opposite direction from the objective one (e.g., greater than 1.0 rather than less than 1.0). For example, if a subject estimates that a barometer's falling is more likely to occur if the weather is going to be dry than if rain is in the offing, then he will undergo an objectively expected decrease in truth possession about the weather by learning that the barometer is falling and conditionalizing on that evidence. This shows that Laudan's optimistic claim about the truth-conducive properties of learning new (credible and) relevant evidence is incorrect. A theorist must provide substantial additional argumentation to substantiate the case that admitting all (credible and) relevant evidence into court is an invariably truth-conducive policy. Laudan provides no such additional argumentation.

The indicated theorem, it must be admitted, relies on a certain measure of degree of truth possession, a measure that not all researchers working in the field would accept.Footnote ³⁸ Different sorts of so-called “scoring rules” are usually preferred. However, although this theorem might not be replicable under different scoring rules, the important point for present purposes is that it is virtually certain that different scoring rules would not allow any similar theorem about expected increases in truth possession that make no qualification about the relationship between a reasoner's subjective likelihoods and the objective likelihoods. One cannot get such a theorem in which the antecedent makes assumptions only about objective likelihoods. Some sort of alignment between subjective likelihoods and objective likelihoods is needed, as the foregoing weather example illustrates. This moral carries over to the criminal trial case. If jurors have very bad estimates of the objective likelihoods relating pieces of evidence and target events (e.g., criminal acts), one cannot expect their receipt of objectively relevant evidence to translate into expected increases in movement toward truth.

Admittedly, Laudan is occasionally sensitive to the point being made here, but not sensitive in a thoroughgoing fashion. At times he seems to think that the risk of juror mishandling of the evidence is restricted to the task of keeping their emotions in check:

Yet the problem at hand is not just a matter of emotions, fairness, or common sense; it is a matter of associating suitable likelihoods with the evidence in question. That depends heavily on past experience and familiarity with relevant situations, including situations that jurors themselves may never have encountered (e.g., being threatened by interrogators to confess or face stiffer charges). These matters are cognitive, not emotional; and they depend on more than mere common sense.

In certain passages Laudan seems to concede the possibility of juror misconstrual of the evidence less narrowly conceived. He therefore offers a “codicil” to his “meta-rule” that relevant evidence should always be admissible:

What is unclear is why the burden of argument for exclusion should be set so high, i.e., at the level of “compellingness.” If there were a proof or a prima facie case that any piece of relevant evidence contributes positively toward recognizing the truth, it would perhaps make sense to treat admissibility as a default that can only be overcome with powerful evidence. But there is no such proof or prima facie case to be had. Correctly appreciating the weight of evidence is a difficult task; it often requires more than mere common sense and is not necessarily sealed by a judge's instructions or opposing counsel's arguments.

Laudan envisions a future in which proposed or existing evidentiary rules will be empirically tested to determine if they contribute to or detract from the goal of truth-seeking. We agree that such testing would be very valuable, but it seems unlikely to occur—at least on a systematic scale. In the meantime, we must make do. Laudan scorns the rules that have developed in the absence of such rigorous testing as mere “folk mythology.”Footnote ⁴¹ But his derision has no real basis. Indeed, one might argue that jurists' decades of experience with the law and the consequences of its rules provide a strong foundation from which to draw conclusions.

Our final point has a moral somewhat analogous to the preceding one. It is the point that an adequate epistemology, including an adequate epistemology of the criminal trial, cannot deal exclusively with objective matters; it also must take subjective factors into account. We next apply this idea to Laudan's treatment of standards of proof and judges' instructions to juries. In his discussion of how to interpret the BARD standard of proof, Laudan argues persuasively against a purely subjective interpretation. Surely some element of the objective must be brought into the picture. However, Laudan goes too far in apparently ignoring the necessity for some subjective factors to be part of an adequate mix.

Laudan critiques the notion that the BARD standard of proof should be interpreted (to jurors) in terms of purely subjective probabilities, or degrees of credence. We have no major disagreement with him here. He finds three “objective” interpretations more interesting and promising.Footnote ⁴² The first formulates BARD in terms of credibility: there is credible inculpatory evidence or testimony that would be very hard to explain if the defendant were innocent, and no credible exculpatory evidence or testimony that would be difficult to explain if the defendant were guilty. The second formulates BARD in terms of plausibility: the prosecutor's story about the crime is plausible, and one cannot conceive of any plausible story that leaves the defendant innocent. The third formulates BARD in terms of reasonability: the facts rule out every reasonable hypothesis you can think of that would leave the defendant innocent. Although Laudan accepts none of these proposals, he clearly leans toward interpretations that invoke what would be credible, plausible, or reasonable to believe—that is, worthy of belief—as opposed to what a juror actually believes. In other words, objective norms of thought need to make an appearance in the standard, not simply what jurors happen to think.

Our chief worry is whether interpretations of these sorts would give jurors sufficient guidance and whether they ultimately escape reliance on subjective facts. Just as one cannot be assured that jurors will make accurate estimates of conditional probabilities, so one cannot be assured that jurors will make accurate assessments of what is “credible,” “plausible,” or “reasonable” in a story or hypothesis. Thus each of the three objective interpretations Laudan considers presents a standard that may be beyond jurors' ability to recognize correctly as applying or failing to apply in the legal case before them. Psychological evidence is widely thought to reveal strong currents of irrationality in naïve thinking. How can we be confident that lay jurors can correctly apply interpretations that invoke rationality or similar normative (epistemic) concepts? Indeed, although the (intended) meanings of credible, plausible, and reasonable differ from thought to be credible, thought to be plausible, and thought to be reasonable, respectively, it is questionable whether the average juror has any way of determining whether the normative terms apply in a given case other than by seeing whether she herself finds various stories and hypotheses credible, plausible, or reasonable. In other words, jurors may have to see how they (subjectively) judge a given hypothesis or possibility in order to apply the normative terms in question. For this reason, it is doubtful whether presentation of these normative terms would enable jurors to get beyond reliance on their own subjective assessments, contrary to Laudan's intention.

In short, although we agree with Laudan that no purely subjective interpretation of BARD is likely to be adequate, we do not think that the task of formulating an adequate interpretation for jurors should ignore the subjective use to which they would put any objectively worded interpretation. Again, this point is analogous to our point about the truth-conduciveness of relevant evidence.

Although we disagree with Laudan on a number of points, we also wish to express our appreciation for the clarity, boldness, and systematicity that he brings to the subject of legal evidence and particularly our agreement on the fundamental point that the rules of evidence and procedure ought to be carefully scrutinized. Many more philosophers, and especially epistemologists, should be bringing the distinctive tools of their trade to this domain. Laudan's important example is well worth following. He also sets a fine precedent by becoming impressively knowledgeable in the criminal law so that he can address the debates with genuine expertise and scholarship as well as theoretical flair.