Printer Friendly

Evidence, procedure, and the upside of cognitive error.

   A. The Correlation Between Knowledge and Interest
   B. The Necessity and Nature of the Test
      1. The oath
      2. The ordeal
      3. Demeanor
      4. The polygraph
      5. Perjury and obstruction
   C. Testing Semantic Content
   D. General Evidence for Consistency and Detail
   A. Comparison of Preparatory Tasks
      1. Story construction
      2. Contingent thinking
      3. Memorization
   B. Additional Legal Applications
      1. Use of writings in preparing testimony
      2. Prior consistent and inconsistent statements
      3. Asymmetric treatment of consistent and inconsistent
      4. Witness exclusion
   C. Summary
   A. Unanticipated Questions
      1. The law of surprise
   B. Closed-Loop Control of Questioning
      1. Implications for discovery
   C. Fatigue
      1. Implications for discovery
   A. Witness Versus Herself
   B. Witness Versus Questioner
      1. The battle over consistency
      2. Anticipation and closed-loop questioning
      3. Cognitive artifacts
      4. Fatigue and tag-team questioning
      5. The sincere witness's strategy
   C. Factfinder Versus Witness Versus Questioner
      1. Accommodation from attorneys
      2. Accommodation from the judge
      3. Cognitive artifacts
   A. The Creation of Cognitive Artifacts
      1. Two case studies
      2. The special efficacy of cognitive artifact evidence
   B. Control and Destruction of Real Evidence
      1. Cognitive artifact evidence
      2. Other real evidence


As an information processor, (1) the human mind falls far short of its own ideal. Working memory--where intermediate information is stored in multistep thinking (2)--holds only about seven items, (3) and only for several seconds. (4) Longer-term memories can be purposefully acquired, but only with substantial effort and attention. (5) Even when successfully acquired, long-term memories cannot always be retrieved on demand. (6) Indeed, focusing the attention required for managing long-term memory--as well as for many other cognitive tasks--can itself be a challenge. (7) And just to make matters worse, the mind's limited capacity further constricts as it quickly grows fatigued. (8)

When legal scholars make reference to the limitations of human cognition, it is usually in connection with bad news about the legal system. Nowhere is this more clearly the case than in applications of cognitive psychology to evidentiary procedure. (9) A vast literature stretching back several decades is almost uniform in its grim prognosis for factfinding. (10) Eyewitnesses, we are told, see little, remember less, and often believe they remember what they never saw. (11) Witnesses' indications of confidence in the accuracy of their perceptions bear little relationship to actual accuracy. (12) Juries and judges nonetheless place substantial weight on witnesses' expressions of certainty and doubt. (13) Indeed, factfinders are purportedly prone to a full catalog of cognitive illusions and biases, symptoms of their general mental deficiency. (14) "Hindsight bias," (15) for example, causes them to believe that accidents that happened to occur were bound to happen and, thus, that sufficiently cautious defendants were negligent. (16) The "representativeness heuristic" (17) induces them to place too much weight on whether the evidence matches their mental picture of, for example, negligent behavior, and too little weight on the base frequency of such behavior in the population. (18)

Though somewhat less pronounced, a similar pessimism informs applications of cognitive psychology to substantive-law fields. To be sure, many of the applications in this more recent research program are put forward primarily as challenges to the methodology of law and economics, with its "working assumption" that individuals are perfectly rational. (19) But many also speak directly to the detrimental effect of cognitive limits in the substantive-law arena. Thus, we read that consumer choice with regard to dangerous products may not be the rational expression of economic preference, but rather the outcome of systematic mistakes in judging risk which have, in turn, been manipulated by producers. (20) And even in the absence of transaction costs (as conventionally defined), parties to a nuisance dispute may not bargain around an ill-informed verdict, as the Coase theorem (21) would optimistically predict. (22)

Taken individually, many elements of the literature applying cognitive psychology to substantive and procedural law make valuable contributions, not just to our understanding of the legal system, but also to its practical design. (23) Viewed as a whole, however, the literature offers a woefully unbalanced account of what human imperfection means for law. Ironically, nowhere is the account more askew than with regard to evidentiary process. For when it comes to how the legal system itself processes information, the limits of the human mind are as much an instrument as an impediment. (24)

Elucidating the instrumental value of cognitive imperfection to evidentiary process is the primary object of this Article. (25) Secondarily, the Article sets out to show how the law of evidentiary procedure tilts the playing field of litigation in a manner that mitigates the cognitive failings of certain actors while exacerbating the failings of others. En route to both objectives, the Article provides a novel perspective on several critical, though underanalyzed, topics in the law of evidence and procedure. More generally, in exploring how cognitive limits bear on the pervasive issue of legal information processing, the Article casts new light on the significance of cognitive limits for law as a whole.

The argument for the existence of an upside to cognitive deficiency begins with a simple, but apparently elusive, point: the normative status of mental limitations depends on the object to which the mind is directed. This Article stands apart in its approach to cognitive error because it focuses on a set of mental objectives that to date have been largely ignored in applications of cognitive psychology to law, both substantive and procedural.

Where cognitive psychology has been applied to substantive legal rules, these rules are most often considered in a form that is disembodied from the problem of how the law gathers the information required for implementation and enforcement. In this artificial realm, we may stipulate, the fact that individuals have limited cognitive capacity is usually a detriment to social welfare. When, for example, the individual's mental objective is to identify remote contractual contingencies or minimize accident avoidance costs, the greater her capacity to process information, the better for society.

Applications of cognitive psychology to evidentiary process, on the other hand, have addressed only certain issues in that field. Near exclusive attention has been paid to the perception and memory of the truly disinterested witness and the rationality of the truly impartial trier of fact. Again, we may stipulate that when the mind under study belongs to one of these beneficent actors, the greater its cognitive capacity, the better for social welfare.

But surely the dramatis personae of evidentiary process consists of more than just disinterested witnesses and impartial factfinders. After all, what about the "bad actor" (26) of evidentiary process--the one who cogitates about evasion, who ruminates on deceit, who deliberates about how to break the law without a trace; the one who lies under oath, (27) alters phone logs, (28) shreds key documents? (29) We well understand the archetype of the evil genius. (30) Then, too, we should understand that when the individual puts her mind to fabricating an airtight alibi, covering up a paper trail, or obfuscating clear of her opponent's burden of proof, her cognitive limitations are a boon for system function. (31)

The perfect mind, after all, effortlessly constructs the perfect lie, anticipating much of cross-examination. It easily comprehends the constellation of evidence that is likely to be available to the other side and weaves its lie through these fixed objects with agility. Even when a question is unanticipated, the perfect mind spontaneously and consistently extrapolates in a manner consistent with both its own prior statements and other available evidence.

Needing no notes to keep straight the details of its illicit plan, the perfect mind leaves no evidentiary trail of documents, records, and other "cognitive artifacts." (32) Even when the perfect mind must resort to making records, it remembers if and where the records are stored, it keeps straight whether and where any copies were made, and it comprehensively erases these evidentiary traces of its activities--without then leaving additional traces of the erasure.

The limited mind, in contrast, has difficulty thinking more than several steps ahead in anticipating the give and take of cross-examination, it pauses to answer the questions it has failed to foresee as it attempts to calculate what kind of answer will be both consistent with prior testimony and also unlikely to lead to yet another difficult question. The limited mind quickly fatigues, whereby its attention and working memory become woefully impaired. Late in the day, deep into the deposition, it cannot remember the details of the answer it gave to a sneakily similar question five hours earlier. Rehearsal is an intense, time-consuming, and doubtful solution. The limited mind, unable to fully anticipate the questions it will face, is equally unable to target preparatory effort. What it ultimately decides to rehearse, it is slow to learn and quick to forget.

In the process of perpetrating its crime, tort, or breach, the limited mind takes copious notes to help it pay attention, to help it comprehend, to help it remember what to do next and what to say to whom. It forgets that these notes exist. It fails to keep straight the genealogical tree by which these notes are duplicated and passed along to exponentiating generations of recipients. If it takes notes about these notes, it loses track of these as well. It, thus, chooses unhappily between exposing itself to the limitations of its unaided ability or exposing itself to discovery of the aids it uses to free itself therefrom.

In terms of the social objectives of evidentiary process, the fact that humans are of limited cognitive capacity has several positive implications ranging along the timeline from ex ante to ex post. Furthest ex post, after evidence has been presented, it means that liars are more frequently caught in their lies. To the extent that one views truth finding per se as an ultimate objective of evidentiary process, this is in itself a positive effect.

Medially ex ante, at the time when parties and witnesses are choosing what evidence to prepare and present, it means that those who might choose to lie if they thought they could get away with it choose instead to be truthful. To the extent that a lie might still have been successful, this too enhances the accuracy of factfinding. And even to the extent that the lie would have been detected anyway, this saves the incremental cost of exposing, prosecuting, and punishing its perpetration.

But perhaps the most important (33) positive effect of cognitive shortcomings resides even farther toward the ex ante end of the timeline. This is the effect on the choices that people make well before the rare prospect of litigation arises, choices they make on a daily basis, on the job, in the conduct of business, in the ordinary course of their personal lives. These are the individuals' "primary-activity" decisions, and they implicate tort law, contract law, commercial law, corporate law, and every other substantive law field.

Faced with the prospect of legal penalties and motivated by the desire to avoid them, the hypothetical perfect mind and the actual limited mind make very different choices in the primary-activity arena. Each chooses between two costly and imperfect means of avoiding penalties. One option is to incur the cost of complying with the substantive law, thus reducing the chance of legal sanction by attacking the possibility of penalty at source. The other course of action is to ignore the costly dictates of the substantive law and focus instead on precluding or erasing evidence of noncompliance. The perfect mind, because it is more likely to succeed at evidentiary evasion at lower cost in terms of money, time, and effort, is more likely to choose cover-up over compliance. The limited mind, to the extent that it at least is cognizant of the fact that it is limited, more likely finds compliance, rather than evasion, the most cost-effective means of avoiding legal penalties.

Thus it is that buried beneath a mountain of scholarship cataloguing and substantiating the foibles of sincere witness memory and factfinder decisionmaking, and drowned out by the dialogue between lawyer-psychologists and lawyer-economists about whether it is appropriate to analyze law as if its subjects were perfectly rational, lies a fundamental and fundamentally distinct point about the implications of cognitive imperfection for law as a whole. On the isthmus of evidentiary procedure--over which every substantive field of law must pass on its way from paper rule to real effect the big story is not that law is hampered by cognitive imperfection, nor even that it has "adapted" (34) to human error, but rather that the law relies upon mental limitations, that it exploits cognitive shortcomings, and that it would not function nearly as well were humans truly perfectly rational.

This Article is organized as follows. Parts I through IV analyze testimonial evidence. Part I discusses the purpose and mechanism of testimony generally, concluding that differences in the cognitive load required for creditable testimony across sincere and insincere witnesses play a crucial role. Parts II and III analyze these differences in cognitive load on a task-by-task basis, Part II focusing on witness preparation and Part III on witness performance. Part IV concludes the discussion of testimony by considering how the playing field of litigation is tilted to exacerbate the limitations of the insincere witness and mitigate the limitations of other actors such as questioners and factfinders. Part V turns to the topic of real evidence.


This Article takes a different view of the role of cognitive limits in testimony because it takes a different view of testimony. Explaining that view, and contrasting it with the view grounding existing analyses, is the purpose of this Part.

An amateur etymology of the word "testify" provides a useful template for both endeavors. A first attempt at derivation turns up two candidate etymologies, each evoking a very different picture of the signified. One traces testi- to tres, meaning "three," and stare, meaning "to stand." (35) In the image of testimony thus drawn, a "third party stands by" (36) to offer her observations of the underlying event or condition, her "thirdness" representing her lack of personal interest and, thus, her lack of incentive to offer anything but the truth.

An alternative, and seemingly more straightforward, etymology links "testimony" to "test," which derives in turn from testum, a clay bowl in which metals were heated to determine their true nature. (37) Here, the party standing by is hardly disinterested in how her offer is received, having much to gain if the metal she tenders is taken as gold. Unable to discern gold from lead at room temperature, the tenderee "'puts" the metal "to the test," heating it until gold's invisible chemistry emits telltale signs. (38)

Were etymological alignment the touchstone for legal analysis, the existing literature on cognitive psychology and witnesses would be well positioned. As it turns out, tres-stare is the accepted derivation, and there is in fact no recognized connection between "testify" and "test." (39) Correspondingly, research on eyewitness fallibility, which constitutes by far the largest wedge in the research pie chart, is self-consciously focused on a species of witness having no personal stake in the outcome of the suit. (40) Similarly, the literature on witness preparation emphasizes above all the danger of corrupting impartial, well-meaning witnesses. (41)

Yet whatever their mutual alignment, both the etymology and the existing literature are out of line with the reality of evidentiary process in its modern manifestation. For the several reasons outlined in this Part, modern testimony is, and has been for some time, more akin to testing metal than tallying third-party accounts. (42) The false etymology is in fact the truer metaphor.

A. The Correlation Between Knowledge and Interest

Were it not for the literature's tight focus on "third parties," it would probably be too obvious to point out that those most often "standing by" to testify are the parties themselves. (43) In the civil arena, it is hard to imagine a witness list or deposition schedule that does not include the plaintiffs and defendants. In criminal suits, although defendants may refuse to testify, (44) what data exist indicate that they often do. (45) And the victims of crime might well be regarded as "parties" for this purpose, given their potential interest in obtaining retribution, constraining an accused competitor, or establishing facts for follow-up suits. (46)

Indeed, in many cases, the parties are the only witnesses on specific elements or defenses. Sex offense suits, for example, are thought to display a frustrating "he said, she said" character on the issue of consent, (47) and this is thought to justify special evidentiary rules. (48) In fact, "he said, she said" seems less a special problem than a general tendency in the generation of evidence. Auto accidents on deserted roads, agreements reached over the phone, (nonsexual) assaults in dark alleys--all of these pit one party's word against another's. Indeed, it stands to reason that perpetrators will go out of their way to commit their crimes in private and false accusers to script their fabrications out of public view.

Nonparty witnesses do often play an important role in factfinding. But such witnesses may themselves have an interest in litigation outcomes. Friends or relatives may be called to support the defendant's alibi. (49) Current employees, their careers in the balance, may testify for their employers. (50) Disgruntled former employees may be keen to blow the whistle. (51) Competitors--business or social--may stand to gain if one side loses. (52) Coconspirators may cooperate with police in return for a favorable deal. (53)

The prevalence of witnesses with evident interests is no coincidence. To be useful in illuminating the underlying event or condition, a witness must have been both present and paying attention. (54) Neither characteristic is randomly determined. To the extent that location is volitional, the physical presence of a witness will tend to indicate that she has a connection to the parties or an interest in what has allegedly transpired. With regard to mental presence, cognitive psychology has long analogized attention to a scarce resource that humans carefully budget. (55) It follows--theoretically and experimentally---that those who "pay" attention tend to have a reason to do so. (56)

All told, then, it should come as no surprise that the defendant's husband was the one who witnessed her comings and goings on the day of the murder. Nor that the person who witnessed the accident from the passenger seat of the plaintiff's car was the plaintiff's son. (57) Nor that the executives of the company are the ones who know the most about the allegedly illicit plan to derail competitors. (58) Nor that only the defendant's partner in crime understands the complicated scheme for insurance fraud. (59) Nor that the defendant was sleeping with the minor daughter of the acquaintance who turned him in. (60) Nature does not draw nonparty witnesses like lottery balls. Rather, the natural choice of witnesses exhibits a severe "selection bias" toward those who care about the outcome of the resulting lawsuit.

Just the opposite selection bias informs the vast experimental literature on cognition and evidence, with its near exclusive focus on witnesses who are disinterested by experimental design. At the same time, the experimental results in this literature are rife with episodes in which disinterested subjects fail to pay attention, (61) suggesting that people remember surprisingly little of what occurs right before their eyes when such events do not concern them personally. The literature interprets this as bad news for evidentiary process. But another interpretation is that these experiments are proving their own irrelevance--that if a witness did pay enough attention to have real information to offer, that witness is probably not like the disinterested subjects of these experiments.

An apologist for the existing literature's tight focus on disinterested witnesses might respond that, despite these tendencies toward partiality, truly disinterested witnesses do make an appearance in evidentiary process. (62) Moreover, when they do, they have a disproportionately powerful impact precisely because they are impartial. (63)

As a preliminary matter, it must be pointed out that the shrunken territory of relevance to which this response retreats is wildly out of proportion to the great estate held by the disinterested witness in existing analysis. But even putting this aside, though it must surely be true that not all witnesses are interested in fact, it is just as obvious that we cannot tell who these witnesses are merely by looking at them. Self-interest is as complicated and unfathomable as the web of relationships that ties society together. And it would be naive to expect the covertly biased witness to help us trace her connections to the suit. Especially so, if we accept the apologist's premise that apparent impartiality magnifies testimony's import.

Indeed, the same dynamic that produces selection bias in favor of self-evidently interested witnesses also suggests that witnesses lacking apparent bias will tend to have hidden interests to the extent that they were present and paying attention--that is, to the extent that they have information to offer. Consequently, the best position is probably to treat all witnesses as if they are interested, even though we know that this is not literally true. In this particular yet practical sense, then, the focus of the cognitive psychology literature on witnesses who are axiomatically impartial is not merely incomplete but completely inapplicable.

Existing scholarship notwithstanding, evidence law has long been suspicious of witnesses' hidden interests. As early as 1836, in the English case of Thomas v. David, (64) it was not even in contention that defense counsel could ask the plaintiff's witness, his "female servant," "whether she did not constantly sleep in the same bed with her master." (65) More recently, the U.S. Supreme Court, in United States v. Abel, (66) explicitly confirmed that impeachment for bias would continue under the then-pubescent Federal Rules of Evidence. The Court upheld a district court's decision to admit evidence that the accused and his witness both belonged to an organization, the "Aryan Brotherhood," (67) whose members pledged to lie on each other's behalf. The Court also carried forward the time-honored rule that the impeaching party may proffer secondary witnesses or nontestimonial evidence for the purpose of showing that the subject witness has an interest in the outcome of the suit. Such "extrinsic evidence" of bias is allowed even though it will often have the effect of lengthening trials and discovery to cover matters not directly related to the historical merits of the case. Other forms of impeachment--by prior inconsistent statements, by contradiction, or by character for untruthfulness--are generally not so favored. (68) In other cases, the Court has gone so far as to locate the accused's right to impeach for bias in the Confrontation Clause of the Sixth Amendment. (69)

B. The Necessity and Nature of the Test

The fact that those who have information to offer also tend to have an axe to grind does more than just call into question the predominant approach. It also suggests that the system is well advised to find a way to work with witnesses despite their interests, extracting from them as much real information as possible. This leads, in turn, to the following question: when a self-interested witness comes forward offering to relate the content of her memory, how can we judge whether that offer is genuine? The pseudo-etymology of "testify," which aptly illustrates our predicament, also suggests a solution. Finding ourselves in the same position as the skeptical tenderee of supposed "gold," we do just as she does. We devise a test.

Any test has three basic elements: a set of unobservable characteristics, features of ultimate interest whose invisibility necessitates the test in the first place; a set of observable test outcomes, which are, by choice, discernible to the tester; and an associative mechanism by which different unobservable characteristics predictably manifest themselves in different observable test outcomes. In the metal-testing metaphor, for example, the unobservable characteristic is the chemical composition of the tendered metal. The observable test outcome concerns how the metal behaves before the naked eye under extreme temperatures. And the principles of thermochemistry provide the associative mechanism linking atomic structure to behavior in the heated bowl.

For the test of testimony, the unobservable characteristic of interest is reasonably clear: the relevant content of the witness's long-term memory. The other test components, however, are not so obvious. Regarding observable test outcomes, we know that these outcomes must revolve around the witness's performance "before" the factfinder. (70) Yet, the kind of performance that should be required, the aspects of that performance that ought to be monitored, and how these aspects ought to be judged all require further analysis. Such analysis must simultaneously identify the associative mechanism. Their association with unobservable characteristics is, after all, the only reason to observe test outcomes.

Before proposing a test design that fits the bill, this Article, in the next several Parts, runs through a number of test designs that fall short. This derogatory review will serve two purposes. First, the failure of these alternative designs will illustrate the necessity of two required attributes for observable test outcomes: (1) correlation with the content of memory, and (2) an adequate cost differential between producing the test outcome when it is truthful and producing it when it must be fabricated. (71) Second, the observation that these alternative designs are collectively deficient will point toward the necessity of another form of test, which a subsequent Part will propose.

1. The oath

Imagine that after swearing the witness in, we ask her whether her memory conforms to her proponent's depiction of events, accepting her answer as true, no further questions asked. In other words, consider the observable test outcome: bare testimony under oath.

Federal Rule of Evidence 603 instructs that the oath be administered in such a way as to "awaken the witness's conscience." (72) (To be sure, it also awakens the perjury statute, a point to which we shall return.) Supposing this to be the case, swearing in produces a kind of signal cost differential that is keyed to the underlying content of the witness's memory. That is, the cost of the rewarded signal (73)--here the moral cost of answering "yes," that her proponent's depiction is accurate is higher if her proponent's depiction is, in fact, inaccurate. In this respect, the oath is at least a step in the right direction. The existence of a differential cost--whether measured in time, money, or moral composure--is a necessary condition for the efficacy of testimony. Without it, the rewarded signal would be completely manipulable, and the social objectives of evidentiary process completely defeated. (74)

To elaborate, were there no signal cost differential, witnesses would face the same testimonial choice problem whether their proponent's depiction was true or false. Their signal choice problems being the same, so too would be their actual signal choices. (75) The test of testimony would, thus, fail to distinguish sincere and insincere accounts, and the goal of truth finding would not be advanced. Neither would the related, but distinct, goal of primary-activity incentive setting. Parties who followed the dictates of the substantive law in the primary activity and told the truth about it in litigation would fare no better in litigation than parties who followed the dictates of their private interests in the primary activity and lied about it in litigation. (76) Accordingly, prospective litigation payoffs would provide no incentive to comply with the substantive law. (77)

But although the oath takes a step in the right direction, that step is probably too short. The requisite cost differential must not only exist but also be comparable in magnitude to the organic incentive that it is meant to counteract. With regard to truth finding, if the litigation payoffs from answering "yes" are larger than the moral cost of lying, those moral costs will not be decisive in testimonial choice. Similarly, a primary-activity actor who realizes that she can acquire sizable primary-activity gains without incurring unfavorable verdicts simply by paying the relatively small conscience cost of lying will not be adequately deterred from the wrongful primary-activity act. (78)

Indeed, viewing the oath in terms of not just the costs, but also the benefits, of lying raises the question of whether it has any positive effect at all. If we accept without analysis that the witness comes to the stand with her conscience asleep, it is perhaps fair to suppose that her sense of consequence is also in repose. The solemnity of the oath may awaken not only her conscience, therefore, but also her awareness that what she says hereafter is not idle chitchat but of immediate moment to her private interest, if there were ever an occasion to bear the moral cost of lying, the oath may be telling her, "This is it."

2. The ordeal

In contrast to sworn corroboration, the rewarded signal in the medieval trial by ordeal (or more precisely, the pedagogical caricature of it discussed here (79)) was difficult to fake. Unfortunately, that signal was also largely unrelated to the underlying characteristic of interest.

In one variant of trial by ordeal, the accused was forced to walk nine paces with a hot iron rod laid across his upturned palms. His hand would then be "covered under seal" (80) for three days. Thereafter, the burn wounds would be inspected to see if they had festered. If they had, the defendant was declared guilty; if they had not, he was acquitted. (81)

Reining in our imagination regarding how the system might have been manipulated (as it were), let us suppose, for purposes of illustration, that whether the defendant's wounds festered depended on biological factors that were effectively out of his control. Thus, the ordeal entailed very large cost differentials in the sense that the defendant could not successfully pretend that the wound had healed. Yet, this sizable cost differential was not correlated with the underlying characteristic of interest: namely, guilt or innocence. Though the ordeal may have been an effective test of the immunologic health of defendants or the sterility of the cleric's bandaging job, it was not an effective test of what it was designed to uncover.

3. Demeanor

Demeanor evidence consists of everything the factfinder can readily discern about a witness's performance, excluding the semantic content of her testimony. (82) Whether the witness fidgets, gesticulates, averts her gaze, whether her voice cracks, stutters, or rises in pitch, how frequently she pauses and for how long--all these are demeanor cues. (83)

Once solemnly regarded, (84) such cues are now generally discounted by evidence scholars. (85) In terms of the framework under consideration, demeanor evidence suffers variously, and sometimes simultaneously, from both a cost differential problem and a correlation problem.

Consider first demeanor's cost differential problem. Suppose that we believe people tend to feel guilty when they lie and, in attempting to dissociate themselves from their deception, avert their gaze. (86) Based on this belief, we might decide to credit a witness's answer only if it is pronounced while looking the questioner in the eye. But even assuming that there is a natural tendency to avert one's gaze while lying, the impulse is too easily masked. Consequently, were it the case that averting one's gaze was recognized as a sign of lying, liars would learn to fix their gaze, and looking away would be a poor signal of insincerity. (87) Consistent with this hypothesis, experimental research is nearly unanimous in finding that gaze aversion is of essentially no use in detecting lies. (88) Such results are typical of demeanor cues generally. A recent meta-analysis (89) by Bella DePaulo and her colleagues, (90) which analyzes experimental data on a list of 158 potential deception cues (91) across a virtually comprehensive list of prior studies, (92) reports that "effect sizes" (93) are "small" (94) for many classic demeanor cues--including fidgeting, blinking, shrugging, face touching, smiling, lip pressing, object fidgeting, brow lowering, and lip-corner pulling. (95)

Where demeanor evidence does not have a cost differential problem, it has a correlation problem. The demeanor cues for which DePaulo and her colleagues find "large effect sizes" consist mainly of cues related to stress and arousal. These researchers find, for example, that liars speak with higher-pitched voices, noting that "pitch typically rises with stress." (96) But the link between stress cues and lying involves a problematic medial inference. Stress cues are cues only to an emotional state, which is in turn posited to be associated with lying. Yet any witness--insincere or sincere--will likely be nervous at trial or on deposition. Stress cues exhibited in court may, therefore, signal only the gravity of the situation, not the insincerity of the witness. Correspondingly, the relative success of stress cues in experimental settings may be an artifact of the low level of baseline stress generally felt by laboratory subjects. (97)

4. The polygraph

The logic behind the polygraph is similar to the logic behind stress-related demeanor cues. Rather than eyeballing ticks, blinks, and scratches, polygraph testing employs various instruments to precisely measure blood pressure. breathing rate, and palmar perspiration. (98) These scientific readings are probably more reliable in measuring stress per se. But the polygraph technology does nothing to shore up the tenuous connection between stress and lying in a high-stress situation. Thus, the polygraph suffers from the same correlation problem as stress-related demeanor cues. (99)

The polygraph also suffers from a significant cost differential problem. The test can be beaten by detecting the questions on which the innocent are meant to exhibit more stress and then simulating stress in answering those questions. Stress may be simulated by pressing one's fingernails into one's palms, biting one's tongue, or pushing one's toes into the floor. Consider, for example, the "Guilty Knowledge" variant of the test. Assume that the perpetrator is aware of the fact that she dropped her hat at the scene of the crime. We ask the suspect the multiple choice question: "What did the perpetrator drop at the scene of the crime: (a) a scarf, (b) a glove, (c) a hat, or (d) a shoe?," instructing the suspect to answer "yes" or "no" after each alternative is named. When the test works as it should, the guilty suspect exhibits more arousal in answering "no" to question (c), while the innocent suspect shows no discernible difference across the alternatives. The guilty suspect can artificially produce the same constancy, however, by pushing her toes into the floor when answering "no" to alternatives (a), (b), and (d). (100) Correspondingly, there is evidence that individuals can be quickly taught to beat the polygraph. (101)

5. Perjury and obstruction

Perhaps we ought to throw away the bible, the hot iron, and the electrodes and simply inform the witness that if she lies, she will serve ten months in prison for perjury or obstruction of justice. (102) That would presumably produce a significant correlation; only those who answer falsely pay the high cost. And it would also produce sizable signal cost differentials; the additional cost of answering "yes" when it is false, compared to when it is true, is time behind bars.

The problem, of course, is that stating a punishment is not the same as figuring out whom to punish. The fact that perjury and obstruction carry stiff sentences does not mitigate the problem that they too require some test to determine who should be forced to serve those sentences. Bare reliance on perjury and obstruction is, thus, a form of "process begging," wherein the problems of evidence are not solved, just shunted back to a different, unanalyzed phase. (103)

C. Testing Semantic Content

These problems with the oath, demeanor, and the polygraph do not by themselves dictate doing away with such devices. (104) Flaws of correlation and cost differential afflict all forms of evidence; and, sensibly, the requirement of relevance is not that evidence be perfect, but only that it contribute some amount of probative value. (105) Nonetheless, the problems we have identified with these candidate test outcomes indicate that they alone are not sufficient for the test of testimony, that there must be something more.

Conspicuously absent from the array of testimonial test outcomes so far considered--and oddly deemphasized in much modern scholarship on evidence (106)--are cues based on the semantic content of testimony. Quite apart from the issue of how the witness says what she says--whether under oath, with averted gaze, or with increased pulse--is the issue of what the witness says.

This Article's analysis of cognitive imperfection rests on the premise that semantic content is an important observable test outcome for testimony. Two dimensions of content are emphasized: detail and consistency. Testimony is consistent when it contradicts neither itself nor external facts that can be established by the opponent. Testimony is detailed to the extent that it is contradictable. Cognitive imperfection enters the picture as a central component of the associative mechanism. Detail and consistency are viable observable test outcomes because producing them in conjunction draws substantially more cognitive load when the witness is insincere than when she is sincere. (107) The imperfection of human cognition is the wedge that props open this differential in difficulty. To substantiate these assertions, Parts II and III review and contrast the cognitive tasks of producing consistent, detailed testimony for the sincere versus the insincere witness.

D. General Evidence for Consistency and Detail

Before turning to these specifics, however, it is worth noting the several forms of empirical evidence that point toward the efficacy and actual use of detail and consistency as observable test outcomes.

In addition to substantiating the weakness of demeanor cues, the meta-analysis conducted by DePaulo and her colleagues provides support for the proposition that detail and consistency are effective cues. Of the twelve deception cues that were based on a reliable number of independent samples and resulted in large effect sizes, a substantial portion concerned cues that may be considered components of detail and consistency. Specifically, the "prediction that liars would provide fewer details than would truth tellers was clearly supported." (108) "Like good novelists, truth tellers sometimes describe the settings of their stories; liars were somewhat less likely to do this ... and they provided nonsignificantly fewer unusual details." (109) Regarding consistency, "the lies made less sense than the truths. They were less plausible ... [and] more likely to be internally discrepant." (110)

Indeed, experimental data is likely to understate the importance of detail and consistency in actual litigation. Many of the experiments DePaulo and her colleagues considered did not, and could not, replicate the kind of relentless probing that one might see at trial or on deposition. It is, thus, likely that liars supplied relatively more detail in the laboratory than they would have in litigation because they knew they would not have to answer follow-up questions. Along these same lines, inconsistencies in the liar's tale that might have been uncovered via additional probing were not recorded by these experiments.

Additional empirical evidence of the efficacy of detail and consistency--albeit of a more casual form--is provided by the actual use of these observable test outcomes in litigation practice. Here we can identify two sources of support. First, two of the five modes of impeachment generally recognized in evidence law directly implicate detail and consistency testing. Thus, a witness's prior inconsistent statements are liberally admitted to impeach her credibility. (111) Furthermore, a witness may be impeached by contradicting her testimony with the prior statements of other witnesses, with documents or other tangible evidence, or by judicial notice. (112)

The view that these two forms of impeachment primarily concern the observable test outcomes of testimony differs from conventional conceptions of their function. The most often stated rationale for these forms of impeachment rests on the (unnecessarily Latin) phrase falsus in uno, falsus in omnibus, an inference logically distinct from that which we have just described. But this classical deduction cannot be the full justification for these two modes of impeachment. If it were, these modes would reduce to a special case of impeachment by showing the witness's character for untruthfulness. After all, falsus in uno, falsus in omnibus is practically the definition of character impeachment by proving specific instances of lying. Yet the terms under which character-based impeachment may proceed--for example, the rules regarding when extrinsic evidence (113) may be offered--are typically distinct from those under which a witness may be impeached by prior inconsistent statements or contradiction. (114) Making sense of this difference in law requires distinguishing between the kind of deductions the factfinder makes based on a specific instance of lying on a past employment application, for example, from the kind of deduction she makes based on a specific instance of contradiction in current testimony. The conventional explanation fails to do so. The distinction offered here is that impeachment by prior inconsistent statements and contradiction are observable test outcomes for the test of testimony, whereas character impeachment generally is a means of calibrating that test to fit the disposition with regard to truthfulness of the particular witness before us. (115)

Lastly, what empirical evidence there is on the frequency with which various modes of impeachment are employed in court suggests the importance of consistency testing. Professor H. Richard Uviller, for instance, finds in his survey of trial judges that prior inconsistent statements and contradiction are the most often used ways of determining credibility. (116) In accord with this finding, other experienced commentators make bold statements (though often not specifically substantiated) about the frequent employment of these two impeachment devices. (117)


True to the literature on cognition and evidence generally, legal scholarship on witness preparation is preoccupied with the disinterested witness. Judge Jerome Frank, probably the most cited commentator on the issue, soft-pedals witness partiality, focusing on the danger that attorneys will cause vaguely sympathetic witnesses to "unconsciously" alter their "honest[] belie[fs]." (118) Professor Stephan Landsman, who has also importantly written on the topic, explicitly assumes away witness interests from the get-go, turning his full attention to the distortive effect of partisan suggestion on sincere witness memory. (119)

The natural correlation between knowledge and interest, as described in Part I, calls into question the external validity of studying witness preparation solely with reference to witnesses who are impartial by hypothesis. But even taken on its own terms, the literature on witness preparation appears to rest on a precarious conception of the relationship between the neutral, well-meaning witness and the zealous, misleading advocate. At times, the literature seems to implicitly assume that the witness will be fooled into thinking that the partisans "preparing" her are similarly impartial, a feat of astounding naivete and one seemingly at odds with the generally jaded attitude toward lawyers. (120) At other times, the literature appears to suggest that the impartial witness will be caught up--albeit "unconsciously"--in a kind of team spirit that makes her want to help win the case for the first attorney who happens to have contacted her--a team spirit that, however ungrounded in personal interest and however spontaneously generated, cannot be reversed or counterbalanced by subsequent contact with the other side.

In contrast to existing treatments of witness preparation, and in line with this Article's general approach, the witnesses who populate the account of preparation herein not only have interests but also view preparation as an occasion to actively pursue them. These witnesses are not fooled into false memory, but rather choose what to transmit to the factfinder. The focus is thus shifted from an account of the vagaries of sincere memory onto the relative difficulty of fabricating memories versus retrieving them.

Always in the background in the analysis to follow are two premises. First, witness preparation cannot simply be eliminated by fiat. Privately conducted, (121) rarely discussed in public, (122) and yet reportedly ubiquitous, (123) witness preparation would be no less difficult to prevent than excessive carelessness, inefficient breach, or criminal activity. Second, witness preparation not only changes the cost of lodging a creditable performance but is itself a cost thereof. Thus, occasions where a party has paid extensive costs to prepare its witnesses and has thereby been able to fool the factfinder into thinking her witness's testimony is sincere ought not to be automatically regarded as instances of system failure. The costs of preparation--because, as we shall see, they are greater for the insincere witness than for the sincere--act as a differential tax on noncompliant behavior.

Consider then the insincere witness's list of things to do in preparing for testimony, and compare this to the analogous list for the sincere.

A. Comparison of Preparatory Tasks

1. Story construction

The first challenge for the insincere witness is to construct a story. In so doing, the insincere witness is caught in the vice of detail and consistency.

Consistency has internal and external components. Testimony is internally consistent when it does not contradict itself The antitrust deponent who asserts in the morning that she has never even met her competitor's CEO and in the afternoon that their relationship is strictly social can expect to have her testimony discounted. The same goes for the suspect whose alibi implies that she was simultaneously in Seattle and Los Angeles.

Maintaining internal consistency is difficult when the constructed narrative is rich. Each added detail sprouts a new set of logical implications, and ensuring that the tendrils do not cross can draw substantial cognitive load. The conjunction of even a comparatively small number of propositions may produce a surprisingly large network of implications, any pair of which, however distantly related, may contradict. The entire field of mathematics, after all, can be built up from a relatively short list of axioms, (124) and mathematicians devote careers to finding as-yet-undiscovered implications.

An externally consistent story does not contradict what the factfinder--aided by an ever-vigilant litigation opponent--can ascertain about the world. (125) If the deponent says he took the train to West Chester on Tuesday at 10:00 a.m., there must have been a train at that time. If the trial witness says she saw the accident as she was coming out of the diner, there must have been an unobstructed view from that vantage point.

In constructing an account that furthers her interests, the liar must scale over, duck under, and steer around a host of external facts concerning the world in which she sets her story. Indeed, this obstacle course must be run largely in the dark. Much of what the other side can use to contradict the liar's account may not become apparent to her until she runs headlong into it at trial or on deposition. Perhaps such obstacles can be illuminated and avoided with enough preparation. But "enough" may be a lot. And, as noted, expensive and time-consuming preparation is also a component of punishment and deterrence. (126)

Conversely, the witness who withholds detail for fear of contradicting either herself or what her opponents can establish about the world is like the student that leaves her bluebook blank so as not to give an incorrect answer, or the figure skater who attempts no difficult jumps so as not to risk a fall. All three see their "scores" suffer for playing it safe. (127) Accordingly, David Boles, commenting on his deposition of Microsoft chairman William H. Gates, reports that he was pleased to see, as the deposition got underway, that Gates was feigning ignorance. "You want to capture as much of that as you can.... If you have a witness that says 'I don't know,' you want to get him to say 'I don't know' a thousand times." (128)

While the insincere witness must mindfully construct a detailed and consistent story from scratch, this task is largely accomplished for the sincere witness by retrieving memories of actual events. Internal and external consistency are built into the sincere witness's account by virtue of the fact that the reported events really did unfold in time and space and, thus, are automatically in accord with the laws of chemistry and physics. If the witness really saw the accident from where she was standing, then there was indeed a sightline from that position. If she really took the train at 10:00 a.m. on Tuesday, then the trains did in fact run at that time.

Memory retrieval, though not free from difficulty, is probably among the easiest of cognitive tasks. The chief source of difficulty is finding "retrieval paths" to the particular location in long-term memory where sought-after information is stored. Once the path is found, memory unfolds in a mysteriously effortless, largely automatic process. (129) Remembering is a highly leveraged task. One bites into the right sort of cake, and volumes of remembrance ensue. (130)

The relative ease of memory retrieval is supported by both existing doctrine and psychological data. Apparent Proust fans, the writers of the classic evidence treatises wax eloquent on the miracle of remembering in justifying the practice of allowing the stumped witness to consult written materials on the stand to help jog her memory. (131) In addition, cognitive psychologists have confirmed the efficacy of a series of common techniques designed to systematically locate retrieval paths and, thus, produce the flourish of relevant memories that Proust intuitively describes. The result can be what is termed "hypermnesia": remembering what was forgotten (as opposed to amnesia: roughly, the forgetting of what was remembered). (132) Time is a key ingredient in the application of these techniques. (133) One waits until the effect of the prior unsuccessful retrieval paths has left working memory so that a new approach has full sway and is not sidetracked. (134) Forgetting is, thus, an important part of remembering, as anyone who has returned to a crossword puzzle knows well.

That said, refreshing the memory is hardly an error-free enterprise. Moreover, there is an important sense in which even the retrieval of sincere memory is also a task of story construction. (135) And pointing out these nuances of memory is certainly a valid exercise in the study of witness testimony. Nevertheless, it is not an exercise that deserves the nearly exclusive attention that it has received in the literature on cognition and evidence. However much sincere memory is subconsciously constructed, this pales in comparison to the degree to which insincere testimony is consciously constructed. However difficult it is to retrieve an automatically consistent story from memory, it is far more difficult to construct a consistent story out of whole cloth.

And let us not forget that the insincere witness also has quite a bit of memory to retrieve. She too must remember what really happened in order to weave her story around the observable manifestations of real events. Even more, she must call up her store of knowledge about the world in order to avoid external inconsistency. Thus, even if the sincere witness does face difficulties in retrieving memories, these largely cancel out with similar difficulties for the insincere. What is left is a differential chiefly occupied by the insincere witness's difficult task of story construction.

2. Contingent thinking

Even were the insincere witness able to construct a detailed and consistent narrative, this would hardly exhaust her list of things to do. Such preparation might produce an impressive performance on direct examination. But the witness can count on the opponent to explore spurs off the main storyline during cross-examination. Given the even greater difficulty of spontaneously extending a detailed and consistent account--a difficulty we explore in more detail in the next Part--the witness will want to do what she can to anticipate antagonistic probing. (136)

The kind of contingent thinking required to do so is even more difficult than it may at first seem. The witness must not only anticipate the questions her account will raise, but also how' her planned answers to these questions will lead to follow-up, second-order questions, as well as how her answers to second-order questions will themselves be followed up with third-order questions, and so on. The number of answers to prepare increases exponentially as we progress through each layer of questioning, (137) and keeping up with this explosion of contingent inquiries seriously taxes the capability of the human mind. A similar form of contingent thinking is required for games like chess, wherein each player must anticipate not just how her opponent will react to her current move, but also how she will counter-react to each of her opponent's possible reactions, as well as how her opponent will counter-counter-react, and so on. Research on chess indicates that even world-class chess players, who we may suppose play chess nearly every waking hour nearly every day, are able to look only several moves ahead. (138) Lying witnesses, who play a far less structured game, usually as a sidebar to their careers, will likely perform far worse on this score.

Such contingent thinking is of much less import for the sincere witness, who may find herself in the position of spontaneous memory retrieval but never spontaneous story construction. This is not to say that the sincere witness, who nonetheless has interests to pursue, could not also benefit from anticipating the questions she is likely to face. Such anticipation will help her to locate memories and to transform those memories into language so as to make her performance more impressive. It will also help her to adorn her answers with favorable nuance and spin. But the insincere witness has to be just as concerned with retrieval, articulateness, and spin. What creates a sizable cost differential is the fact that the insincere witness must also worry about fabricated content.

3. Memorization

Witnesses must generally testify from memory. This requirement is implicitly enforced by limits placed on both the witness's use of notes while testifying and the questioner's use of leading questions on direct examination. (139) Conventional explanations for the general requirement as so implemented are unsatisfying. We read that the use of notes is discouraged in order to ensure that the factfinder hears the witness's testimony in the witness's own words. (140) This fails to explain why it is any more or less the witness's own words if she is reading from her own notes. Similarly, leading questions are discouraged supposedly to prevent the advocate from putting words into the witness's mouth. But why would the sponsoring party wait until the witness is on the stand to put words into her mouth? (141) That is what "witness coaching" is for--at least according to the same literature. (142)

The requirement of memorization is all the more puzzling in the context of the existing literature's focus on the foibles of sincere eyewitness memory. If we may assume that we have before us a witness who is both disinterested and of severely limited memory, why should we rely on what that witness happens to be able to recall right here, right now? Why not let her read from notes--notes that she may well have recorded while the event was flesh in her mind, notes that she has had a chance to contemplate, question, and revise?

Perhaps a better source of explanation lies in the fact that in organic litigation, as opposed to laboratory experiments of our own design, we are far from certain that the witness is sincere, and we believe implicitly that one way to tell the liar from the truth teller is to exploit the cognitive shortcomings that make lying a markedly more difficult task than truth telling. Accordingly, we recognize that testimony from memory, while it certainly hampers the sincere witness, hampers the insincere even more.

In attempting to construct a detailed, consistent, and robust story, the insincere witness engages, in the first instance, her "working memory." (143) Somewhat like RAM on a computer, working memory is relatively easy to access but small in capacity and quick to dissipate. (144) Thus, as she settles on an account, the insincere witness must also be transferring it from working memory to long-term memory, which is more difficult to access but more permanent. (145) Although this transfer happens repeatedly in daily life, it is relatively difficult to control, requiring time and attention. (146) In contrast, the sincere witness need not memorize what is already in memory. And though she may benefit from memorizing phrasing and emphasis, the same is true for the insincere witness.

One may object to this account of memorization by pointing out that the ancient doctrine of "present recollection refreshed" does, after all, allow the witness to refer to notes in order to refresh her memory while testifying. (147) But the law charges a price for this allowance, a price which bears more heavily on the insincere. If the witness refreshes her memory with a writing, the other side may inspect the writing, cross-examine the witness thereon, and introduce relevant portions into evidence. (148) The effective result is a targeted prohibition on refreshing memory with records that one would not want one's opponent to see. This is likely to help the sincere witness more than the insincere, a point which we shall return to immediately below in discussing similar disclosure roles for writings used in preparing for testimony.

B. Additional Legal Applications

1. Use of writings in preparing testimony (149)

Notes would be of special assistance to the insincere witness in preparing her testimony, lowering for her the cost of providing a detailed, consistent, and robust account. The witness could graph out the main storyline and all its spurs. And for purposes of memorizing what she thereby produces, notes would fix content for repeated rehearsal.

Yet the legal landscape governing the use of notes in preparing for testimony has two prominent features that limit the extent to which the insincere witness can make use of this cognitive aid. Both of these features serve to make the cognitive task of preparation more difficult for the insincere witness without also making it more difficult to a comparable degree for the sincere.

a. Embarrassing notes

As noted, when the witness uses notes to "refresh her memory" while testifying, such notes may be viewed by the opposing attorney, who may ask the witness about them and potentially have them admitted into evidence. (150) Codifying an incipient shift in the common law, (151) Federal Rule of Evidence 612 extended this treatment to writings used to refresh memory before testifying (152) when such treatment is in the "interests of justice." (153)

In the often-cited case of James Julian, Inc. v. Raytheon Co., (154) for instance, Raytheon learned that counsel for Julian had prepared a briefing binder for use by Julian's principals, officers, and employees in preparing for their respective depositions. (155) Julian's witnesses did not make use of the binder during the deposition, (156) but testified from memory. Raytheon argued that this binder should be disclosed under Federal Rule of Evidence 612. The court agreed. Indeed, it did so after first finding that the binder was opinion work product. (157) Federal Rule of Evidence 612 was held to trump the usual protections for such materials. (158) Other courts have found similarly with regard to attorney-client privilege. (159)

Even broader disclosure rights apply to materials used solely in preparation by expert witnesses. (160) Federal Rule of Evidence 705, for example, states that a testifying "expert may ... be required to disclose the underlying facts or data [upon which her opinions or inferences are based] on cross-examination." (161) And Federal Rule of Civil Procedure 30(c) makes clear that this rule applies to depositions in civil cases. The upshot is that testifying experts in civil cases are routinely required to disclose any notes they may have made in writing their required reports (162) or preparing for depositions and trial testimony. This includes early, and perhaps less favorable, drafts of their reports and even scribbles in the margins of consulted references.

Such rules give the opponent the right to see a preparatory writing and the option to offer that item into evidence. The opponent's exercise of her right to see the writing has real effect only if seeing the writing reveals to her something that helps her assemble her case. Similarly, the opponent's possession of an option to offer the item into evidence has real effect only if she exercises the option, which she will do only if admitting the item into evidence helps her case. Therefore, the rules have greatest effect when the witness's preparatory writings are damaging to her proponent. This, in turn, implies that these rules are less of a threat to the sincere than to the insincere. The sincere witness is far more likely to prepare with writings that are not only not damaging but possibly even helpful to her proponent's case--writings like hearsay correspondence or reports that corroborate her story. (163) The preparatory writings of the insincere--more precisely, the preparatory writings that the insincere would like to draft and compile, and would do so but for the risk that they would turn up in her opponent's file or in the jury room--are likely to be of a different flavor. Her notes might show trial and error in graphing out a storyline. Her preparatory binder might include the potentially damaging information around which she was forced to craft her story.

b. Oral preparation

The disclosure provisions in Federal Rule of Evidence 612 apply only to writings, (164) not to the content of oral preparation. Oral preparation generally retains opinion work product protection. (165) The fact that the law favors oral over written preparation is likely to induce witnesses and their proponents to do the same. (166) Professor John Applegate, concerned that oral preparation has greater potential for error, sees this substitution as a perverse and harmful effect. (167) Yet in light of the differing cognitive tasks faced by the sincere and insincere witness, it gains substantial justification. Oral preparation is differentially beneficial to the sincere witness. The witness who seeks only to jog her true memory can often make do with purely oral preparation. The witness who must construct a story from scratch, play out as many of its logical implications as possible, keep track of what the other side can prove about the world, and memorize it all to boot could make good use of a war room, let alone a notepad.

2. Prior consistent and inconsistent statements

In United States v. Microsoft, the government alleged that Microsoft met in June 1995 with its rival Netscape to present a plan to divide up the Internet browser market, with the inducement that Microsoft would invest in Netscape. When Microsoft chair William H. Gates was asked about this on deposition, he stated only that a subordinate had broached the possibility of Microsoft's investing in Netscape but that he "didn't see that as something that made sense." (168) The government then introduced an e-mail from Gates sent prior to the June 1995 meeting in which Gates proposed attempting to collude with Netscape, using Microsoft's investment to coax Netscape to accept the arrangement. (169)

The contradiction of statements made outside of current testimony raises both internal and external consistency issues. Prior out-of-court statements present external consistency issues to the extent that we view the witness's prior statement as fixed. As such, preparing around one's own prior statements is similar in character to preparing around external facts established by the statements of other witnesses. (170) To the extent that the witness is unaware of a prior statement, it can be used to discredit her current inconsistent testimony. To the extent that she is aware of it, it will bound her current testimony by virtue of its potential impeachment value. (171)

Prior statements are also to some extent endogenous, the choice of what to say outside of court being part of the same preparatory task as the choice of what to say at trial or on deposition. (172) Here prior statements raise internal consistency issues. Like preparation on the eve of trial, such protolitigation activities are cognitively daunting. Indeed, given the necessity of looking farther forward in time, the presence of greater uncertainty as to what evidence will end up being decisive in litigation, (173) the typically longer time span over which this kind of "preparation" must be consistently kept up, and the greater necessity of multitasking with one's daily activity, this form of preparation is likely to be even more difficult. (174) In fact, it seems probable that individuals are rarely successful at avoiding the creation of all potentially damaging statements as they go about their daily activities. As most cases enter litigation, the field is likely strewn with prior statements that either effectively bound the witness's current testimony or may be used to discredit it. The fact that prior statements--outside current testimony--are a chief mode of impeachment points in this direction. (175)

3. Asymmetric treatment of consistent and inconsistent statements

A more specific aspect of the rules concerning prior statements is illuminated by viewing such statements as part of the task of preparing for litigation.

It is fairly clear that the importance of remembering not to make potentially damaging statements outside of court is enforced by liberal (176) rules for impeachment by prior inconsistent statements. But what about the importance of remembering to make corroborating statements? One's first impression might be that such affirmative pronouncements would fall under the proponent's admission of prior consistent statements. Thus, perhaps the witness could buttress her testimony by pointing to times in the past when she had said precisely the same thing as she is now saying on the stand. Nevertheless, in the wake of the Supreme Court's decision in United States v. Tome, (177) the trend has been to strengthen the restrictions on the admission of prior consistent statements so as to largely prevent this kind of deployment. (178)

In analyzing this trend, it is important to keep in mind that such restrictions do not imply that there is little sense under current law in remembering to make corroborating prior statements. It is well established that prior silence may be regarded as inconsistent with affirmative statements made later during testimony. (179) "[I]f the former statement fails to mention a material circumstance presently testified to, which it would have been natural to mention in the prior statement, the prior statement is sufficiently inconsistent." (180)

What could justify allowing impeachment by prior unnatural omission, while broadly disallowing corroboration by prior consistent statements? The answers implicate, again, the law's exploitation of cognitive shortcomings. It is easy to remember to make a corroborating statement at some point. It is more difficult to constantly keep in mind the necessity of making corroborating statements whenever a situation arises where one would have done so.

4. Witness exclusion

Federal Rule of Evidence 615 provides that the court may exclude nonparty, nonessential (181) witnesses from trial "so that they cannot hear the testimony of other witnesses." (182) Given exclusionary rules of this nature, the preparatory exercise of contingent thinking is all the more imperative when a set of insincere witnesses testifies in cahoots.

Nearly insurmountable consistency problems arise with respect to the external consistency of each witness when members of the group are individually asked a question that the group did not anticipate and whose answer the group, therefore, did not coordinate. (183) A single insincere witness has at least some chance of coming up with a consistent answer to an unanticipated question. But given the infinity of consistent answers that might be chosen, even if all witnesses answer consistently, the possibility that two or more witnesses will choose the same consistent answer is extremely small.

Anticipation is not so imperative for a group of sincere witnesses. Their answers to unanticipated questions will tend to be automatically consistent by virtue of the fact that they all described what actually occurred. Again, this is not to say that they will not have differing opinions and perceptions. It is only to say that discrepancies will be far less likely to arise, far less severe, and far easier to explain than those produced by the on-the-spot answers of a group of witnesses that have failed to fully coordinate their fabricated account.

The fact that the necessity of anticipation is multiplied along with the number of witnesses is perhaps part of the reason why corroborated testimony is typically given greater weight than the testimony of a single witness. (184) The alternative explanation, merely analogizing greater numbers of witnesses to larger sample sizes, is inapt. It founders on the fact that insincere witnesses can coordinate their preparation. Given this, the sample of witness testimony would not be independent were all questions anticipated, and a greater sample size would just be more of the same. Only when it is recognized that cognitive bounds prevent the group of witnesses from anticipating all the questions they may face does the greater weight accorded to corroborated testimony make sense. Thus, the logic of witness exclusion is also intimately connected with the exploitation of cognitive limitations.

C. Summary

Preparing for testimony can be difficult and costly for any witness. But for the insincere witness of bounded cognitive capacity, the amount of time and effort required to prepare consistent, detailed, and robust testimony is particularly daunting. In fact, no amount of rehearsal could fully prepare the witness for all contingencies, which makes performance--as opposed to preparation--an independent challenge, one that we take up in the next Part.


Much of the scholarship and case law on the performance of testimony labors in the shadow of the classic assertion that the law has a "preference for live testimony." (185) But precisely what is it about being "live" that makes it so important? Can't a well-rehearsed soliloquy be performed "live"? Is a spontaneous free-for-all still "live" if it is prerecorded? As questions like these indicate, "live" is a multifaceted concept, and the importance of any one of its facets is far from self-explanatory. Before translating the law's amorphous "preference" into solid policy, therefore, the concept of "live" must be deconstructed and its vital co-ingredients identified.

Existing attempts to give content to the law's "preference for live testimony" commonly fall back on two subsidiary tenets. The first is the importance of observing witness demeanor. As we have seen, this is a questionable proposition. (186) And in any event, it is far from clear how observing demeanor relates to "live." Couldn't demeanor be recorded for playback? Why, for that matter, couldn't the declarant's demeanor be reported by the hearsay witness along with what the declarant asserted? Were the declarant's demeanor somehow a substantive issue in the case, we would certainly rely on a witness's personal knowledge thereof. How can there be less justification for such reliance when demeanor plays a less central role in the case?

The second tenet typically proffered to justify the law's preference for "live" testimony is the efficacy of cross-examination. De rigueur here is quotation of Wigmore's superlative--perhaps the most quoted phrase in evidentiary process--that cross-examination is "the greatest legal engine ever invented for the discovery of truth." (187) But like "the law's preference for live testimony," this phrase too is analytically unsatisfying on closer inspection. (188) First, even if we accept cross-examination's efficacy on faith, precisely how it implicates "live testimony" is again inadequately explained. Is there something about cross-examination that could not be accomplished by multiple rounds of interrogatories crossing in the mail?

Second, and relatedly, the question of what cross-examination is and how it works remains essentially unexamined. Resort to Wigmore's soundbite does little to bring the concept into focus. To assert that cross-examination is "the greatest legal engine ever invented for the discovery of truth" is not to explain how that engine works, or even to inventory its moving parts. Perhaps it is no coincidence that Wigmore's assertion first appeared in 1904, (189) seven years after the invention of the gasoline-powered automobile (the greatest combustion engine ever invented for the transportation of humanity?), two years after such automobiles were first mass produced, and four years before the first Model-T rolled off the line. Given what the word newly signified, to call something an "engine" at the dawn of the twentieth century was likely to accord it power (a word Wigmore uses three times in this same paragraph) (190) and also mystery. (191) Correspondingly, amazement without much understanding is what Wigmore inspires, for there is little in his multivolume treatise to elucidate what makes this legal engine go. (192)

Precisely one century later, it is perhaps time we looked under the hood. (193) In that spirit, the present Part investigates some of the mechanics of the performance of testimony, inclusive of cross-examination. The focus is on how these institutions run off the difference in cognitive load faced by sincere and insincere witnesses. As such, this Part identifies three specific features of testimonial performance that exploit this differential load: unanticipated questions, closed loop questioning, and fatigue.

A. Unanticipated Questions

The challenges for the insincere witness in maintaining internal consistency are many times multiplied when her first cognitive encounter with a question is at trial or on deposition. (194) The first set of challenges revolves around maintaining internal consistency with prepared testimony.

To begin, consider the first time that the witness encounters an unanticipated question. Likely the witness's "Plan A" will be to find an answer that fits into her prepared story. As her mind runs through possible responses, she must both unfold their salient implications and search her memory for potential conflicts with the story she has prepared. A computer would be able to search quickly through its memory to check consistency with what has been or will be said. The human witness, however, possesses a far less impressive search engine. And even a computer would have difficulty with the spontaneous creativity necessitated by the task of adding additional fabrication to an already fabricated story.

Should no such answer come to mind, the witness will likely resort to "Plan B." This is to spontaneously alter her prepared story to fit the answers that she has been able to think of. Plan B requires even more of the witness, which is why it is Plan B. In addition to remembering her prepared story, the witness must also keep straight what parts of that story she has already told and what she has yet to tell, the former being effectively unalterable. Moreover, going forward, she must remember the alterations. New memories of these alterations will compete with older, presumably more well-established, memories of her prepared story. Confusion is likely to ensue.

As if dealing with the first anticipated question were not hard enough, consider now the witness's challenge in responding to the nth unanticipated question. Here the witness must maintain consistency not only with her prepared story but also with her prior answers to unanticipated questions 1 through n-1. This in turn means that, as the questioning proceeds, the insincere witness must not only be spontaneously creating answers and vetting them against the story of record, but also spontaneously transferring those answers from her working memory to her long-term memory--a task that is difficult when one has time to rehearse, and yet more difficult on the fly.

At the same time that the witness searches through what she herself has said or plans to say, she must also be calling to mind the set of external facts provable by the other side. More than this, if the witness gives her answer on deposition or is subject to being recalled to the stand, the other side will have the specific opportunity to investigate and follow up. She is, therefore, well advised to search her entire store of knowledge in an effort to avoid contradicting not just what her opponents already know, but also what they can feasibly discover. This, in turn, implies that the size of the witness's knowledge store, as opposed to her ability to retrieve from that store on the spot, is also an important cognitive bound. If the witness is fabricating a new part of her story that involves taking a train along a particular route, for example, she must know when those trains run.

For the sincere witness, on the other hand, answering an unanticipated question is at base a single task: memory retrieval. There is no need to remember prepared testimony or past answers to other unanticipated questions. There is no need to instantaneously memorize the current answer for future reference. There is no need to have a broad knowledge base at the ready. Each question may be attacked individually. Consistency with other answers and with the world is automatically ensured by sticking with what is actually in memory. Again, this is not to say that spontaneous memory retrieval is always easy; only that it is a good deal easier than the set of tasks facing the insincere witness in the same position.

That unanticipated questions are associated with a large differential in difficulty as between sincere and insincere witnesses begins to give some content to what we mean by, and what is important about, the "live" presence of the witness. "Live," in this particular sense, means potentially unrehearsed.

1. The law of surprise

One response to this analysis of unanticipated questions is to challenge the importance it places on the element of surprise. Some might say that our system, having evolved beyond Perry Masonic protoprocess, has learned to loathe "surprise," eschewing "the old sporting theory of justice [for] a more enlightened policy of putting the cards on the table, so to speak, and keeping surprise tactics down to a minimum." (195) Yet a peek behind this veil of civility reveals that our system still thrives on catching witnesses off guard. (196)

Consider that "surprise" is definitively not one of the items in the comparatively detailed list of grounds for excluding relevant evidence under Federal Rule of Evidence 403--the catchall provision allowing the exclusion of evidence found to be more prejudicial than probative, and the place where "unfair surprise" would be generally proscribed. (197) Nor does surprise appear in a similar place in the evidentiary codes of forty-nine states. (198) This is not for lack of suggestion and advocacy. Such a prohibition does appear in like provisions of several model codes and was explicitly and systematically advocated at the time of the adoption of Federal Rule of Evidence 403. (199) It was also explicitly and systematically rejected. (200) In the interim, courts and commentators have continued to pay heed to this clear legislative history. (201)

Consider also Federal Rule of Evidence 613, governing the procedure by which a witness may be examined concerning her prior statements, including those of which she is presently unaware. Federal Rule of Evidence 613 abolished certain ritualistic aspects of impeachment by prior inconsistent statement, "useless impediment[s]" to cross-examination, (202) according to the advisory committee, which had the effect of reducing the element of surprise. (203) For instance, under the old rule in The Queen's Case, (204) the cross-examiner was required to give the witness a moment to examine her prior statement in writing before questioning her about it. That rule is replaced in Federal Rule of Evidence 613(a) by the weaker requirement that the written statement need only be simultaneously disclosed to opposing counsel out of sight of the witness.

a. Civil surprise

One counterresponse is to claim that proscribing surprise at trial is superfluous anyway, at least in civil cases, given extensive discovery. (205) This assertion is easy to exaggerate. First, many civil cases have little or no discovery. (206) Second, much of what would produce surprise in a witness--including interviews with other witnesses, notes thereon, and investigative reports prepared in anticipation of litigation--would be protected as work product and so be discoverable only in exceptional circumstances, if at all. (207)

Third, and most importantly, even if discovery reduces surprise at trial, (208) it does little to reduce surprise in discovery. The issue is not whether there is surprise at trial per se. The issue is whether witnesses are surprised at some point at which their forced, unguarded responses form part of the basis upon which future rulings are made. True, such unguarded responses may not, by virtue of hearsay rules, be admissible for the truth of the matter asserted. (209) Even so, they will constrain or dilute later answers by their potential or actual impeachment use. (210) And in any event, deposition transcripts are generally directly admissible at trial for their substantive content if the witness says something inconsistent at trial and is available for cross-examination regarding the prior statement. (211) Indeed, if the witness is a party, all her prior statements, whether or not made on deposition, are admissible against her at trial. (212) Moreover, the court will regard a witness's deposition transcript as an indication of her prospective trial testimony for the purpose of ruling on summary judgment, (213) an increasingly utilized exit along the highway to trial. (214) And of course, the anticipated impact of deposition testimony on summary judgment and trial will be reflected back onto settlement negotiations. (215)

b. Criminal surprise

In criminal cases, police questioning plays a role similar in this respect to that of the civil deposition. Although interrogation may clue the witness in as to how she will be examined at trial, the witness's spontaneous reaction to surprise during interrogation often bounds, or even counts as, trial testimony. Assuming conformity to the procedural requirements of Miranda v. Arizona and its (largely erosive) progeny, (216) statements made to police by the accused are substantively admissible against her. (217) Even in cases where the requirements for substantive use are lacking, the accused's in-custody statements (or silence) may' often be used to impeach her trial testimony, (218) thus gaining indirect substantive effect by constraining what the accused may credibly assert at trial. For witnesses other than the accused, this impeachment use--with its indirect substantive effect--is similarly available. In fact, nonparty witnesses enjoy none of the Miranda protections accorded to the accused. (219)

To be sure, under the "Brady rule," prosecutors have a constitutional obligation to turn over exculpatory evidence to the accused upon request. (220) Further, court rules allow criminal discovery beyond the dictates of this constitutional requirement. (221) But as with civil discovery, although these rules limit surprise at trial, they do little to limit the importance of surprise for trial. They do not, in particular, affect the trial use of statements made to police before any exculpatory evidence is turned over. Thus, the insincere perpetrator, who may not even yet be a suspect, may be forced to craft her alibi before knowing what exculpatory evidence she can use to corroborate her story.

B. Closed-Loop Control of Questioning (222)

While the importance of unanticipated questions rationalizes and defines the "live" presence of the witness, it does little to justify the "live" presence of the questioner. The questioner could always choose and record her questions in secret and then have these read in seriatim to the witness. Some of these questions would be unanticipated, and spontaneous answers would be required. In fact, roughly speaking, such a device exists: the deposition upon written questions. (223) Tellingly, it is rarely used.

The live presence of the questioner is in part explained by the efficacy of closed-loop control in the choice of questions. Under closed-loop control, the questioner uses the witness's prior answers as feedback in choosing the next question. Under open-loop control, in contrast, the questioner would choose her questions and their sequence ahead of time, once and for all. (224)

The questioner's closed-loop control of questioning exacerbates the cognitive difficulties faced by the insincere witness. Consider in this regard how much worse off the insincere witness is than the student who has to prepare for a conventional examination: an example of open-loop questioning. The conventional exam is not specifically adjusted on the run based on the student's prior answers in an effort to probe the topics that the student does not know well. But this is essentially the predicament of the witness. If a deponent, for example, appears to have a good story about a certain component of the case, the questioner moves on to other items, continuing to probe and retreat until she finds what appears to be a weakness. As a result, the witness, unlike the student, cannot play the odds that the examination will cover the material she knows well. To match the student's odds of success, the witness has to cover the waterfront, learning it all in depth, which in tuna increases the cognitive load of preparation.

1. Implications for discovery

Combining this discussion of closed-loop questioning with the earlier discussion of unanticipated questions leads to the conclusion that discovery--both civil (225) and criminal (226)--is a two-edged sword with respect to the instrumental use of cognitive limitations. On the one hand, discovery forces the witness to commit to a story at a time when further investigation of that story is still feasible for the other side. Indeed, it aids such investigation. Therefore, it widens the closed loop of questioning and increases the difficulty of falsely providing consistent and detailed testimony. On the other hand, it may also reduce surprise. When, for example, a party must respond to production requests (227) before deposing her opponent's witness, she may tip her hand with regard to what she could use to contradict the witness's testimony. This in turn reduces the cost of producing successful insincere testimony.

Thus, in reforming discovery generally, or even just scheduling it in a particular case, (228) perhaps the most effective way to make use of cognitive limitations is to divide its function into phased components. First, there is the target-fixing phase, accomplished by encouraged admissions (229) and preliminary depositions (230) of parties and initial witnesses. Next, there is the investigative phase, accomplished by requests for production (231) and the deposition of second-order witnesses. Last, perhaps, is the consistency-testing phase, whereby apparent inconsistencies are tested and made apparent to the other side by means of follow-up depositions or interrogatories. (232) If warranted, the cycle may be repeated. (233)

C. Fatigue

To spontaneity, feedback, and follow-up must be added fatigue as a crucial device for specially aggravating the difficulties faced by insincere witnesses. Part of what is imperfect about the information processor that is the human mind is the fact that its battery is fairly quick to run down. Certain cognitive tasks that it could easily accomplish when fresh become enormously difficult when tired. Mistakes are made. Pretenses erode.

Consider the witness in the seventh continuous hour of her deposition. The lawyer who opened the questioning in the morning takes over again, refreshed. (234) He begins by asking the witness questions, consistent answers to which would require her to remember how she answered other questions posed hours earlier. (235) If those prior questions were unanticipated and her answers thereby spontaneous, remembering those answers is likely to be especially difficult. All the more so if the prior question was posed three hours into the deposition, when the witness was already fatigued and therefore hampered in transferring her answer from working to long-term memory.

Don't all witnesses fatigue, sincere and insincere alike? Certainly they do. But just as different computer operations wear down a laptop battery more or less quickly, mental fatigue impacts different cognitive tasks to differing extents. The ability to remember one's name and phone number, for instance, is unlikely to be seriously affected. More generally, performance on tasks involving only the retrieval of long-term memory remains relatively high. Tasks involving working memory, on the other hand, suffer more greatly. (236) Doing mathematical calculations in one's head, for example, becomes increasingly difficult, because one is less and less able to perform the kind of on-the-spot storage and retrieval necessary for the task. On the continuum with remembering one's address on the one end, and remembering the digit in the ten's place as one carries the two on the other, the sincere witness's task of retrieving actual memories sits closer to the former, and the insincere witness's task of spontaneously vetting and then memorizing fabricated answers lies closer to the latter.

The role of fatigue points to a third important facet of "liveness": a lack of intermittency, as when a scene is produced onstage versus on film.

1. Implications for discovery

Fatigue's differential impact also has implications for discovery reform. The recent trend in civil procedure has been toward limiting the length of depositions. (237) This certainly has benefits. It prevents an arms race of escalating litigation expenses, and it is merciful to deponents. But we should also be mindful of its costs, one important component of which is that shortening depositions allows less latitude for fatigue and, therefore, hampers the successful exploitation of that cognitive limitation in evidentiary process. (238)


The primary object of this Article is to shift scholarship's inertial focus on the downside of cognitive limits by pointing out the ways in which the law of evidentiary procedure exploits, rather than suffers from, such limits. Yet even this wider focus does not encompass the complete picture of cognitive imperfection and law. The full story concerns a competition of imperfections, one that takes place in several arenas.

In the first arena, the mind of the insincere competes against the minds of those who attempt to take advantage of its shortcomings. Within this arena, two separate competitions unfold. First, "meta-actors"--advisory committees, judicial conferences, legislatures, and lawmaking courts compete against the potentially insincere by striving to design a system that reliably foils, and so discourages, deceptive behavior. Second, on-the-ground participants--attorneys, investigators, and interrogators--compete against the actually insincere in attempting to expose her deceit. To recognize these competitions is to acknowledge a fundamental (though not inescapable) predicament: all we have to exploit the limitations of the imperfect mind are other imperfect minds.

In the second arena, the imperfections of the insincere actor vie with the well-documented imperfections of sincere witnesses. (239) Every measure designed to make lying more difficult must not also increase the difficulty of truth telling by even more. And every measure that uncovers more liars must not also mistakenly "uncover" as many or more truth tellers, whose cognitive capacities are equally limited.

In the third arena, the cognitive limits of the insincere compete with those of factfinders. Exploiting cognitive limits to distinguish good information from bad works only if the factfinder is not itself so limited as to miss the distinction.

The fourth and final arena is internal to the potentially deceptive actor. The positive ex ante effects of cognitive limitations require that the potentially deceptive actor should not be so imperfect that she fails to recognize the magnitude of her own shortcomings and, consequently, fails to comprehend that successful evidentiary evasion is of high cost. A too-irrational actor not only fails to deceive others, but also deceives herself into thinking that she will easily succeed.

If all of these cognitive imperfections compete against each other, what justifies the claim that cognitive imperfection is on balance beneficial to evidentiary process? The answer lies in the fact that the system is purposefully, not randomly, structured. The playing field of litigation can thus be tilted to dull the effect of harmful limitations and amplify the effect of beneficial. Indeed, the manner in which the law does just this is a largely untapped source of explanatory and prescriptive power in evidentiary process, one that we explore in this Part.

Two of the aforementioned competitions require no additional attention here. An analysis of the competition between system designers and potentially deceptive witnesses is coincident with an analysis of how the law is and should be structured--the topic of discussion throughout this Article. Similarly, the question of how the law tilts the playing field in a way that favors the sincere witness over the insincere has been the subject of the previous two Parts of the Article.

That leaves three issues regarding the contest of testimony: witness versus questioner, witness versus factfinder, and witness versus herself.

A. Witness Versus Herself

Ideally, potential witnesses and parties have the cognitive wherewithal to understand that fabrication is an ill-fated plan. Witnesses, therefore, opt to provide sincere, rather than insincere, testimony. And primary-activity actors choose compliance and sincerity rather than noncompliance and cover-up.

Understanding the difficulty of fabricating testimony draws far less cognitive load than actually fabricating it successfully. One can be smart enough not to bet one's house on a game of chess against a computer and yet not smart enough to win the game. The substantial distance between these two levels of difficulty makes it plausible and consistent to imagine that most actors' abilities fall somewhere between the greater capacity required to beat the system at acceptable expense and the lesser capacity required to understand that trying to do so is likely futile.

Indeed, there is much direct evidence that we do possess the ability to understand our limits. Whenever we say, "I better write that down," we are in fact remembering that our memories are inadequate. Whenever we say, "Let me think about that," we are thinking quickly enough to know that we think slowly. More generally, our pervasive use of cognitive artifacts in everyday life--our calculators, calendars, and computers--provides strong evidence that we know our limits.

Of course, understanding our own shortcomings is only half of the equation. It must also be clear to us that the system is designed so that successful fabrication tends to exceed those limits. Yet it is important to point out that this hardly requires intimacy with the details of system design. Indeed, the knowledge required is of an extremely general nature. The potential witness need not have mastered the intricacies of impeachment, hearsay, or character evidence. She need not even comprehend the roughest outlines of these provisions. Litigation for her may be entirely a black box. Just as the chess player can understand that the computer will beat her and yet have no understanding of how the program works, the potential witness need not know anything about how litigation will beat her, just that it will. Looked at from this perspective, the informational requirements we are considering appear to be among the least demanding across all of law. If one is willing to accept that tort or contract law has any effect on primary-activity behavior, one should not balk at the informational requirements for the ex ante effects of exploiting cognitive error in evidentiary process.

Even if some witnesses do fail to comprehend the difficulty of successful fabrication, all is not lost. For such witnesses may not succeed. In fact, one might make the stronger argument that they are less likely to be capable of succeeding, that witnesses who are not smart enough to understand that it is difficult to lie successfully are often not smart enough to lie successfully. To the extent that this is so, the fact that some witnesses fail to see the difficulty of lying is less of a problem for truth finding. And incentive setting suffers less as well, since the primary-activity actor who contemplates noncompliance may be discouraged by the relative rarity of witnesses who are both ready and able.

B. Witness Versus Questioner

In considering the contest between witness and questioner, the exploitation of cognitive shortcomings may appear to be nothing more than a tug-of-war. On this account, the cognitive pulling strength required from the witness for successful testimony is just that of opposing counsel, the test's efficacy thereby being limited to situations in which the attorney happens to have greater pulling power. But this assumes that the contest goes off on level ground. In fact, several aspects of evidentiary process ensure that the insincere witness pulls uphill while opposing counsel pulls down.

1. The battle over consistency

Consider, to start, just how unfair the contest is over external consistency. The witness has a "first-mover disadvantage." (240) In the initial encounter between questioner and witness--be that via interrogatory, deposition, interrogation, or interview--the witness answers the questions posed by the questioner with little sense of what the questioner knows and can prove about the world. And even though the questioner's contradicting knowledge may become apparent as he begins to probe potential external contradictions, this revelation comes too late to help the witness. The blind move that the witness has already made is largely irreversible, recorded as it is in the record for ready use in impeaching the witness should she deviate later on.

The questioner's opportunity for subsequent consideration and investigation makes the asymmetry even more severe. The witness having essentially committed to her story, the questioner takes the time to ponder the witness's answers, processing them consciously and subconsciously. She tacks flash cards to the wall, wakes up with new ideas, takes enlightening showers. She seeks out other potential witnesses that are likely to have information about each turn in the witness's account. She visits each location where key events allegedly occurred. Her target fixed, her deadline distant, she systematically and efficiently aims her investigative efforts at the now-revealed coordinates of the witness's story.

A similar asymmetry characterizes the contest over internal consistency, especially when the witness faces a question she has not anticipated. While a deponent, having just taken in an unexpected question, must instantaneously review her mental transcript of prior testimony, opposing counsel, having taken in a witness's unexpected answer, may review the actual transcript, usually with ample time to search for inconsistency.

2. Anticipation and closed-loop questioning

The questioner's ability to make use of closed-loop control in choosing questions not only exacerbates the difficulties faced by the insincere witness, as described in Part III.B, but also greatly simplifies the questioner's task.

To be sure, in a world of perfectly rational actors there is no real difference between closed-loop and open-loop control. (241) Thus, feedback is of no utility for the ideal chess player who can lay out a fully contingent strategy, setting out instructions for how to move after every possible partial history of play. (242) The player does not even need to be there. Like the programmer of Deep Blue, she can leave the instructions with someone (or something) else and later find out the result. Similarly, the ideal questioner could construct one massive set of contingent interrogatories. Every question but the first would begin with "if-then" clauses: "If you answered X to question n then skip to question m; if you answered Y,...."

Of course, this ideal plane is not the playing field of actual litigation. Litigation proceeds in a world that is very much informed by the limitations of the human mind, including, as already discussed in Part II.A.2, the difficulty of contingent thinking. Such difficulties imply not only that the witness will face unanticipated questions, but that the questioner will face unanticipated answers. To this extent, closed-loop questioning accommodates the cognitive limitations of the questioner, who has a second chance to deal with unexpected contingencies as they actually arise.

3. Cognitive artifacts

a. Work product versus writings for witness preparation

Contrast the law's wariness of notes used by witnesses in answering questions, as discussed in Parts II.A.3 and II.B.1, with the law's protective attitude toward notes used by representatives of the opposing side in framing questions. (243) The questioner's notes are not only immune from opponent disclosure rights, such as arise under rules like Federal Rule of Evidence 612, (244) but are indeed specifically protected as "work product." (245) Thus, when the witness graphs out a plan for answering the questions of opposing counsel, she may well be required to turn this over to opposing counsel. If it is damaging enough to the witness's credibility, the opponent may press to have it placed before the jury. But when opposing counsel graphs out a plan for questioning the witness, this is protected from disclosure by a near absolute ban. (246) Says Professor Michael Tigar, "I know a lawyer named Michael Kennedy. He plans cross-examination by making a decision tree. His notes mark a beginning point. If the witness zigs in a certain way, Kennedy will follow with a certain form of counterattack." (247)

Absent waiver, Kennedy's notes are for his eyes only. But if the witness he is questioning makes a similar "decision tree" in planning how to react to Kennedy's questions, this might well end up in Kennedy's hands. Aside from making hay of any choice tidbits, Kennedy could be expected to pointedly question--before the factfinder--what use a truthful witness would have for such a complicated "decision tree" when a truthful witness's simple, blanket decision rule, applicable to any question she receives, would be to "tell the whole truth and nothing but the truth."

b. Litigation software

Recent years have seen the emergence of a vibrant market in litigation software, including such products as Summation (248) and Trial Director. (249) These programs provide counsel with an efficient means of "searching, organizing, and analyzing testimony[,] documents and other disparate information underlying a case." (250) Suppose, posits the Summation online demo, "[y]ou want to prepare for the deposition of Mr. Merkin by reviewing prior testimony as well as previously produced documents." (251) You simply search the term "Merkin" and up pops every mention of his name in all transcripts of prior depositions or cases and in every scanned letter, memo, or e-mail message obtained in discovery or located in your own files. Similarly, if in the midst of his deposition Merkin mentions the Millwater deal, you simply have the associate at your side search "Millwater within ten words of Merkin" and conference with you sotto voce after scanning the results.

In contrast, as we saw in Parts II.A.3 and II.B.1, the witness is largely prevented from making effective use of such software. In the first place, in the midst of testimony or deposition the witness is generally required to answer solely from memory, without the aid of writings, recordings, or computer software. And while the witness and her proponent can, of course, use litigation software to prepare for testimony, the resulting data files may, in the court's discretion, be subject to disclosure and potential admission. (252)

The asymmetry as between questioners and witnesses with regard to the use of litigation software was especially evident in the case against Microsoft:
   [W]hen the videotape rolled in court and ... Gates ... testified
   that he knew nothing of a Microsoft "hit team" to attack IBM, the
   Justice team was there with just the right E-mail, culled from
   millions of pages of documents, to refute his testimony.... [These
   were] results that Justice[] ... got from [its] database package
   ... that helped [it] piece together the most damaging bits of
   video. (253)

Doubtless, Microsoft and its chairman know more than most about how software might be used to testimonial advantage. Conversely, the Justice Department, a ponderous governmental entity, is not necessarily known for its software savvy. And yet--other weaknesses in its case notwithstanding--Justice had the upper hand in the contest of testimony. The disparity across the two parties in their knowledge of how to use litigation software was overwhelmed by the differing extent to which each could employ the tool.

4. Fatigue and tag-team questioning

When a tired witness begins to forget her prepared story or finds it increasingly difficult to field unanticipated questions, there is no one to whom she can pass the baton. It is just her memory that she is reporting, and she is generally expected to soldier through.

But should the questioner sense that exhaustion is beginning to dull her comprehension and creativity, she merely turns the task over to a well-rested colleague. In fact, with some well-honed teamwork, she need not even get to this point. The attorneys can divide up the tasks and rotate on and off throughout the sixteen hours of deposition. Each attorney can take an hour to probe the witness on a particular issue, then have several hours to recuperate and ponder whatever might have been unexpected about the witness's testimony before returning fresh to the witness, who may at this point remember little of what she said during their last interaction.

Thus, while the witness runs a marathon, opposing counsel run a relay. In this way, fatigue is allowed to operate on the witness, separating the sincere from the insincere, while not also affecting the questioners who, in order to fully exploit fatigue, must not themselves be as subject to it.

5. The sincere witness's strategy

Sincere witnesses are also to some extent hurt by the tilt of the playing field in favor of questioners. The competitive advantages given to the questioner are likely to increase the chance that truthful witnesses will be discredited. But merely pointing out this possibility is analytically incomplete. No beneficial effect is without negative byproduct. The issue is whether the benefits outweigh the detriments.

Here this is likely to be the case. The advantages accorded to questioners apply largely when the higher-order cognitive processes of questioner and witness are pitted against each other. When the witness opts to tell the truth, a task which in the main does not implicate the same high level of cognitive processing, she effectively opts out of that competition. Reporting what she actually remembers, she is not engaged in second-guessing what the questioner knows or can later find out about the world, she is undaunted by the questioner's ability to go back and check the transcript, she is happy to have the questioner probe more deeply into any of her answers. Her chief cognitive task is merely to retrieve truthful answers from long-term memory, and if she does this, detail and consistency will fall into place.

Thus, so long as the witness retains a tight grasp on the fixed object of truthful memory, the questioner will have serious difficulty pulling her across the line in their tug of war, regardless of the fact that the questioner pulls downhill. The cognitive advantage of being able to anchor on true memory outweighs the other cognitive advantages accorded to the questioner, as just described.

C. Factfinder Versus Witness Versus Questioner

A spectator may understand that White is in checkmate and yet not understand the clash of strategies that brought the game to that point. Similarly, a factfinder may understand that the witness could not have been in two places at once and yet have no understanding of, nor ability to engage in, the undercurrent of second-guessing and cognitive maneuvering that brought that contradiction to the surface. Thus, the cognitive limitations of the factfinder are not implicated to the same degree as those of the witness, or even the questioner.

In addition, a number of specific features of evidentiary process serve to mitigate the impact of factfinder imperfection.

1. Accommodation from attorneys

In discerning whether testimony lacks detail or consistency, the factfinder has in the opposing attorney something of an ally. If she is good at what she does, the opposing attorney is well aware of the factfinder's cognitive limitations and fully ready to accommodate them.

Professor Thomas Mauet's leading guide for trial lawyers contains a chapter on the psychology of persuasion, a fair portion of which is devoted, not to misleading jurors per se, but rather to making them understand the basics of one's case in light of their limited cognitive capacity. (254) The advocate, says Mauet, "must employ strategies that quickly and easily help jurors understand," "must make the trial vivid and visual," and "must be efficient, move the story forward, and make ... points quickly before boredom sets in and jurors tune out." (255) Mauet notes that jurors, accustomed to television and film, "expect to get everything quickly, in simple, digestible sound bites.... Anything less and you've violated the 'boring rule,' and jurors will quickly change channels." (256) Thus, when the losing lawyer complains, "The jury just didn't understand the case[,]' [t]hat is a lawyer problem, not a jury problem." (257) These ideas are repeated throughout Mauet's guide. For example, in his discussion of closing arguments, Mauet advises returning to the memorable words and phrases that encapsulate the case, (258) using exhibits and visual aids, (259) and generally understanding the limits of one's audience:
   [Y]our closing argument must be efficient. Keep in mind that most
   persons can maintain a high level of attention for only 15 to 20
   minutes. Therefore, your argument cannot overload the jury.
   Instead, it should focus on the themes, the key evidence, and the
   law, and it should strip away the peripheral information. Key
   ideas should be repeated, since repetition is so important for
   retention. (260)

Compare this to the plight of the witness in the midst of her testimonial performance. The questioner not only fails to accommodate the witness's inability to remain focused for more than fifteen or twenty minutes, but uses this inability against the witness. Exhaustion is the object, not the pitfall. Questions are repeated, not key ideas. And the point of repetition is not to help the witness remember, but to expose inconsistencies in a falsely memorized account.

2. Accommodation from the judge

When the factfinder is specifically a jury, it is to some extent aided by the presiding judge. In many jurisdictions, following closing arguments, the judge may summarize the evidence in instructing the jury. (261) The federal system, moreover, goes beyond this in allowing the judge, to some extent, also to comment on the weight of evidence and the credibility of witnesses. (262) One influential treatise maintains that a federal district judge may even provide such commentary as trial proceeds, rather than having to wait until after the parties have rested. (263) Case law support for so extending the power to comment backward in time from the point of instruction is not entirely airtight. (264) But the extension forward in time, from instructions to jury deliberations, is clearly supported:
   After the jury retires, if members of the jury, ask the judge for
   an explanation of some part of the evidence or to hear part of the
   testimony over again, the judge has discretion to grant or deny
   their request. Trial judges rarely deny such requests. In fact,
   when it appears that the jury is confused over an important
   element of the case, it may be reversible error for a judge to
   refuse their request for further information. (265)

3. Cognitive artifacts

Like questioners, and unlike witnesses, juries benefit from the use of cognitive artifacts. Whether juries may take notes while trial proceeds is, in most jurisdictions, left up to the discretion of the trial judge. (266) Many judges do indeed permit it, (267) and many jurors apparently avail themselves of the opportunity. (268) The arguments for permitting juror note-taking invoke precisely the kind of cognitive benefits that we have seen are generally denied to the witness. "The obvious and strongest argument in favor of allowing [juror] note-taking is that, when done properly, it is a valuable method of refreshing memory. In addition, note-taking may help focus jurors' concentration on the proceedings and help prevent their attention from wandering." (269)

Furthermore, while deliberating, the jury, of course, has access to all the cognitive artifacts that have been admitted into evidence. A review of common hearsay exceptions is enough to indicate that these could be quite helpful as cognitive aids. Such exceptions encompass: records of regularly conducted activity (including memoranda, reports, records, or data compilation in any form of acts, events, opinions, or diagnoses); (270) public records and reports (including, in some cases, interpretive reports); (271) records of vital statistics; (272) market reports and commercial publications; (273) and learned treatises. (274) Other more general hearsay "exceptions," such as party admissions (275) and prior statements of a witness, (276) might result in the admission of correspondence and deposition transcripts. Should the jury wish to have some portion of a witness's testimony read back to them, this also is within the discretion of the trial judge and is often allowed. (277) Indeed, it is within the judge's discretion to supply the jury with a copy of the pleadings or indictment, even though these are not evidence. (278)


To this point the Article has focused on the upside of cognitive error with regard to testimonial evidence. But nontestimonial evidence--i.e., "real evidence"--also gains efficacy by virtue of the limits of human reason and memory. This Part of the Article considers two respects in which this is true. First, the cognitive artifacts that individuals use in everyday life to bolster unaided cognition often end up generating important evidence. Second, clearing the evidentiary environment of these cognitive artifacts, as well as other "evidentiary emissions" not related to cognition, is a daunting task for the limited mind--one itself requiring yet more cognitive artifacts, which are themselves potential evidence.

A. The Creation of Cognitive Artifacts

Consider how we litter the world with devices that compensate for our limited cognitive capacity. We make packing lists lest we fail to bring along important items. We make to-do lists lest we fail to do what must be done. We memorialize communications with others to remember and prevent misunderstanding. We buy how-to manuals and save instruction booklets because we cannot figure out how things work on our own. We dog-ear pages because we will not remember where the information we need resides. We make notes in the margins because we will not remember precisely how the information applies to our needs. We use maps because we cannot keep the lay of the land in our heads. We circle important locations and trace out the best routes so we find them easily on the next unfolding. We draw diagrams and charts of complicated buildings because we cannot form and retain a mental picture of the facades, elevations, and mechanicals. We scratch out complicated calculations because it is difficult to remember what digit is in the tens place while carrying the two to the top of the thousands column. We supplement our memories with cabinets and hard drives full of files. We back up electronic data because if we lost key documents we would not remember their contents. We label the backups so we can find them later.

Largely unnecessary for a hypothetical creature with unlimited cognitive capacity, such devices are essential for the actual creatures that we are. Such devices are as well the stuff of evidence. They are the tangible manifestations of thoughts and deliberations that, formulated in the private enclosure of the mind, would emit little or no evidence.

Consider, for example, the following two cases, which we will carry throughout this Part.

1. Two case studies

a. The Friday Night Bank Robber

So far are we from being capable of conducting our affairs entirely in our heads that even the most cognitively proficient among us may eventually be caught by the crutches they employ. Carl Gugasian, (279) who depending on one's definition may be the most "successful" bank robber in U.S. history, (280) was recently convicted precisely by these means. (281) A bachelor's degree in electrical engineering; a master's degree in systems analysis; doctoral work in statistics and probability; (282) native cleverness; (283) a special agility with numbers; (284) training in weapons, survival, self-defense, and map reading with Army Special Forces; (285) uncommon mental and physical discipline (286)--none of this changed the fact that Gugasian could not keep it all in his head. Instead, he needed to draft and store "detailed surveillance notes" (287) on the banks he intended to rob, needed to obtain and keep topographical maps to plan surveillance and escape, (288) and needed to keep instructions about how to clean the firearms he used in his robberies. (289) And even though he was smart enough to hide these firearms--as well as his various masks, and even his surveillance notes--in more than a dozen bunkers spread throughout the woods of northeastern Pennsylvania, he had to draw diagrams of those woods (290) to help him relocate the stores. (291) These cognitive artifacts were the clues that the FBI used to link him to multiple bank robberies after two teenagers stumbled upon his gateway bunker, the one closest to his home, containing among other things diagrams leading to his other bunkers and eventually to him. Fittingly, his arrest, based in part on these cognitive artifacts, occurred outside the Philadelphia Free Library, where he was headed to photocopy yet more topographical maps as part of his planning for future robberies.

b. Martha Stewart

Events surrounding the recent perjury and obstruction convictions of Martha Stewart and her stockbroker Peter Bacanovic provide another salient example of the evidentiary role of cognitive artifacts. (292) As 2002 came to a close, the biopharmaceutical company ImClone Systems Inc. was betting heavily on FDA approval of a new cancer drug. After ImClone's application failed at a preliminary stage in the FDA review process--but before this fact was publicly announced--ImClone's CEO began dumping shares through his brokerage firm, Merrill Lynch. (293) Bacanovic, a Merrill Lynch employee, got wind of these sell orders and informed his client, Stewart, who sold her own ImClone holdings on the eve of the public announcement of FDA rejection. (294)

Two important pieces of evidence against Stewart and Bacanovic are artifacts of their otherwise prudent choice not to rely on fallible human memory. One is an entry made by Stewart's personal assistant in a computer phone log the day before ImClone announced the FDA's negative decision. The entry reads, "Peter Bacanovic thinks ImClone is going to start trading downward." (295) This record corroborates the prosecution's theory that Bacanovic called, left a message, and had his assistant at Merrill Lynch fill Stewart in when she returned the call a few hours later. More than this, the phone log boxed in the range of alternative stories that Stewart and Bacanovic could have devised. (296)

The second source of cognitive artifact evidence directly contradicted the story they did come up with. This was that Stewart had instructed Bacanovic to sell ImClone if it ever fell below $60 per share. This meshes well with some of the other external facts. At the time of the phone message, ImClone was trading slightly above $61, (297) and Stewart, who was about to go on vacation, would plausibly not have wanted to be bothered when it hit $60 precisely. However, the story did not jibe well with Bacanovic's worksheets. A broker to many clients with many holdings, Bacanovic quite naturally kept a printed worksheet laying out current and planned actions with regard to Stewart's holdings. (298) Bacanovic also made notes in pen in the margins of this printed worksheet. Prosecutors made hay of the fact that neither the worksheet nor its marginal notes contained any (genuine) indication of Stewart's instruction to sell at $60. (299)

For our purposes, it is worth asking the obvious question: why did Stewart keep a phone log? Against the baseline of perfect cognition, the main reason must be that this phone log compensates for the limited capacity of human memory: it compensates for her assistant's memory limitations in the time before Stewart gets the message (300) and for Stewart's own memory limitations in the days, weeks, and months to follow. Similarly, Bacanovic kept worksheets because he understood the limits of his own mind. If Bacanovic could have kept in his head the ever-changing current holdings, ever-changing prices, and complicated sell orders of his long and ever-changing list of clients, he would not have kept a worksheet, nor would he have had any need to make handwritten notes in the margins of any document.

2. The special efficacy of cognitive artifact evidence

The Stewart and Gugasian examples also illustrate those aspects of cognitive artifact evidence that make it so useful to the opposition and the factfinder.

First, cognitive artifact evidence is relatively easy to interpret. Certainly, when an individual writes notes to another person, these have to convey information about an event or condition in a way that the recipient of the note can understand. And if the event or condition is communicated to the recipient, presumably it is also communicated to anyone else who uses the same language. Similarly, when an individual writes notes to her future self, she must account for the possibility that her future self may no longer have access to her current memory. Thus, it is almost as if she too is communicating with another individual. We have all had the experience, as students, of returning to our own undecipherable class notes. Despite the required exertion, we eventually learn to take notes almost as if we are writing them to someone else so that when the someone else that is us during reading period returns to them, we can readily understand what they mean without counting on recourse to whatever memory we may have of the lecture from months before. As such, our class notes would probably be of some help to a third party who wanted to know or corroborate other evidence about the earlier class session.

Thus from Gugasian's gateway bunker the police extracted a note describing the location of a bank that had been robbed near closing time on a Friday night. Near the name and location of the bank was the notation "F-7." (301) The bank closed at 7 p.m. on Fridays. Similarly, the hand-drawn maps left in the gateway bunker were intelligible enough to lead police to Gugasian's other bunkers in the woods of Pennsylvania.

Of course, an individual can encrypt her notes. But then she has to remember the key. If the key is simple, the code will be broken. If the key is complicated, remembering it may require yet another set of notes. Perhaps these second-order notes themselves can be encrypted with a second-order code. But this also would have to be remembered, and if not easily deciphered, might require a third-order set of notes. We could continue this regress further. But it cannot be infinite. Just as Gugasian had to have a gateway bunker, eventually there will have to be backstop notes in the "natural language," a language that the individual is confident not to forget because it is refreshed on a daily basis as she communicates with others.

The second efficacious feature of cognitive artifact evidence is its self-multiplying character. In their efforts to cross-reference, store, and communicate, humans have devised myriad ways to accurately and cheaply reproduce cognitive artifacts. Phone logs are printed out several times, copies are distributed, copies are made of the copies. Worksheets are backed up, e-mailed, printed from the e-mail, photocopied from the printout, scanned back into electronic form, and e-mailed out again. As a result, many essentially identical versions of the same artifact may be in circulation, though just one of them will suffice for evidence. (302)

B. Control and Destruction of Real Evidence

1. Cognitive artifact evidence

A month after Bacanovic left his phone message and shortly after the FBI asked to speak with her, Stewart went into her assistant's computer phone log and changed the entry from "Peter Bacanovic thinks ImClone is going to start trading downward" to "Peter Bacanovic re imClone." (303) Similarly, sometime after Stewart sold her ImClone shares and after learning that the FBI wanted to see the relevant worksheets for Stewart's account, Bacanovic found a pen of matching color and made the after-the-fact note "@ 60" in the margin near the entry for ImClone. (304)

Yet, perhaps worried about backups or untrackable hard copies, Stewart later allegedly instructed her assistant to return the phone message to its original wording. (305) Moreover, the prosecution introduced evidence that the ink used to mark "@ 60" was "scientifically distinguishable" from the ink used to make the other notes in the margins of Bacanovic's worksheet. (306)

With more foresight, Bacanovic's and Stewart's evidentiary environment might not have been so cluttered with unfavorable cognitive artifacts to begin with, nor would their efforts at cleanup have been so unavailing. Bacanovic might have thought to ask Stewart's assistant that this particular message not be recorded or that it be recorded in a particularly erasable way. Bacanovic might have come up with the sell-at-60 story earlier on and noted "@ 60" just before or after calling Stewart with the same pen around the same time as the other notations. Stewart might have instituted a program of "routinely" deleting after two weeks' time all phone messages not specifically marked for saving. (307) Or she might simply have specifically deleted the message in question before learning of the FBI investigation, thus potentially avoiding obstruction charges. (308)

The problem, as already encountered in Part II's discussion of contingent thinking, is that humans are not always good at foresight. This is especially so with regard to cognitive artifacts of activities that are themselves complex, like running an unrelated household-products business that requires one's full capacity for prospective thinking, or simultaneously dealing with many clients and many stocks with constantly shifting prices.

Our everyday actions spew evidentiary emissions at every turn, most of which are effectively invisible unless we are specifically paying attention to them. And these emissions do not just sit there in a puddle. As noted, they self-replicate, and often in unpredictable and untraceable ways. By the time one finally has the opportunity to focus on the cognitive artifacts of the noncompliant act, those artifacts may have already propagated well beyond one's practical control. Effectively eliminating cognitive artifacts may, therefore, be as cognitively challenging as working without them in the first place.

2. Other real evidence

There are, of course, many other forms of real evidence aside from cognitive artifacts. But the story for these evidentiary emissions is similar with regard to prevention and cleanup. (309) Minding all the evidentiary emissions of a wrongful activity, while also perpetrating the central activity itself, can be cognitively challenging. Both memory and knowledge are implicated. Among many other aspects of the crime, the murderer must remember to check the car for hair and fiber. And the murderer must have adequate knowledge about science and police work to understand what sorts of things might be used to tie her to the crime.


For most jurists and legal scholars, the fact that individuals are limited in their capacity to process information is a pure negative. It subtracts from their ability to avoid accidents at minimal cost, to contract around remote contingencies, to perceive and recall events, to give evidence the weight it deserves. What these commentators have largely failed to notice is that individuals' deficiencies in information processing also subtract from their ability to foil the legal system in its own task of processing information. And to subtract from a subtracter is to add.

Recognizing the double negative of cognitive limits casts new light on several fundamental features of evidentiary procedure. Consider, for example, the law's posture toward cognitive artifacts--a posture that crosscuts many of the topics discussed in this Article. Certainly, the law understands that the human mind relies on notes, calendars, memos, contact lists, software packages, and the like to extend the reach of its limited ability. Such artifacts are often an actual or potential source of evidence in litigation, as discussed in Part V. And yet, even though the law expects to benefit from access to the speaker's notes from her presentation to the corporate board, it typically will not allow the same speaker to read from notes when she "presents" to the factfinder from the witness stand or the deposition table. Part II explains this general prohibition as an attempt to make the cognitive task of presenting creditable testimony more difficult for the insincere than for the sincere. The sincere witness retrieves what is already in memory. The content of her testimony tends to be consistent by virtue of its occurrence in real time. The insincere must construct her false story around what is likely known about the world and then commit that story to memory. Both witnesses would benefit from access to cognitive artifacts while testifying, but much more so the insincere witness whose task draws greater cognitive load.

Even when the law allows the use of notes--either to refresh memory while testifying or in preparing for testimony--the price it exacts for this leniency weighs more heavily on the insincere. The opponent gains access to the cognitive aid and is given the option to admit it into evidence, an option that will be exercised when exposing the aid detracts from the witness's testimony. The result, in effect, is a prohibition on cognitive artifacts that belie the witness's testimony, a prohibition that has less impact on witnesses who tell the truth.

At the same time that the law strips the witness of the cognitive tools that would make feasible the difficult mental task of providing creditable, insincere testimony, it goes out of its way to protect use of the same kind of tools by the individuals who are largely responsible for creating that difficulty, namely opposing counsel. While the witness's notes, when permitted at all, are laid bare, the questioner's notes are specially protected as work product. While software packages to aid the witness in forming an answer are definitively prohibited, software packages to aid the examiner in forming the question are allowed as a matter of course. By this asymmetric treatment of cognitive artifacts, the law helps ensure that questioners' own mental imperfections do not prevent them from systematically exploiting the mental imperfections of insincere witnesses.

These points about cognitive artifacts provide but a few examples of how recasting cognitive limits as instruments, rather than obstacles, brings to the fore central aspects of system design that have languished largely unnoticed in the gray background of evidentiary process. But an expanded view of cognitive limitations also helps to provide much-needed background structure to topics that have always been at the forefront of discourse. The fact that it is especially difficult for the insincere witness to answer consistently on the fly justifies requiring witnesses to provide immediate answers to potentially unanticipated questions. The fact that humans tire and that fatigue has a greater impact on the higher cognitive requirements of insincere testimony explains why the law typically conducts questioning in continuous session. The ability of the questioner to adjust her line of questioning on the spot based on what the witness appears to have failed to prepare goes a long way towards explaining why it is important that the examiner be able to pose her questions with the benefit of real-time feedback from witness answers. All told, the manner in which these features differentially impede the insincere witness lends both content and justification to the premium that the law places on "live" testimony. Moreover, recognizing the interaction between these features and human cognitive limitations brings us a step closer to answering that question about cross-examination that Wigmore once begged: what makes this "greatest legal engine" go?

All of these points about the positive role of cognitive shortcomings reside within the field of evidentiary process. Yet it is important to note that their impact extends far beyond the topical boundaries of that hybrid field. Without information, the substantive law would be limited to rules that apply uniformly, or perhaps randomly, across all individuals and circumstances. Unable to differentiate among varying events and conditions, it would be largely ineffectual. Thus, evidentiary process--essentially the study of how the law differentiates--is not restricted to a particular subset of human endeavor, like tort or contract or property, but rather ranges across the full span of the legal system. The same is therefore true of the beneficial role that cognitive limitations play in the context of evidentiary process. The utility of mental shortcomings is not an exception to a general rule or a special quirk of a specific field, but an essential feature of the law and an integral part of what cognitive psychology has to say about the legal system.

(1.) See generally DONALD E. BROADBENT, PERCEPTION AND COMMUNICATION (1958) (pioneering the information-processing approach to human cognition with its broad analogy to computer processing).

(2.) See generally ALAN D. BADDELEY, WORKING MEMORY (1986) (describing the working memory model and reviewing supporting evidence). When multiplying 63 by 96, for example, one stores "6 x 63 = 378" in working memory, while calculating 9 times 63. More generally, working memory is central to learning, id. at 41-44, memory retrieval, id. at 44-53, language comprehension, id. at 54-64, and reasoning, id. at 64-72. For a sampling of ongoing research on working memory, see Lisa Feldman Barrett, Michele M. Tugade & Randall W. Engle, Individual Differences in Working Memory Capacity and Dual-Process Theories of the Mind, 130 PSCYHOL. BULL. 553 (2004) (investigating individual differences in working memory capacity); David Caplan & Gloria Waters, Working Memory and Connectionist Models of Parsing: A Reply to MacDonald and Christiansen (2002), 109 PSYCHOL. REV. 66 (2002) (discussing the role of working memory in language processing); Marcel Adam Just & Sashank Varma, A Hybrid Architecture for Working Memory: Reply to MacDonald and Christiansen (2002), 109 PSYCHOL. REV. 55 (2002) (same); Maryellen C. MacDonald & Morten H. Christiansen, Reassessing Working Memory: Comment on Just and Carpenter (1992) and Waters and Caplan (1996), 109 PSYCHOL. REV. 35 (2002) (same).

(3.) See, e.g., George A. Miller, The Magical Number Seven Plus or Minus Two: Some Limits on Our Capacity for Processing Information, 63 PSYCHOL. REV. 81, 91 (1956) (reviewing findings on the number of "chunks" that working memory can hold, emphasizing that the individual determines what constitutes a "chunk"); see also Alan D. Baddeley, The Magical Number Seven: Still Magic Alter All These Years?, 101 PSYCHOL. REV. 353, 353 (1994) (discussing the lasting impact of Miller, supra).

(4.) See, e.g., Murray Glanzer & Anita R. Cunitz, Two Storage Mechanisms in Free Recall, 5 J. VERBAL LEARNING & VERBAL BEHAV. 351 (1966) (finding that the "recency effect"--wherein subjects who read a list better recall the last few items, which are still in working memory--dissipates with a brief, filled delay between hearing and the recall task).

(5.) See, e.g., Fergus I.M. Craik & Robert S. Lockhart, Levels of Processing: A Framework for Memory Research, 11 J. VERBAL LEARNING & VERBAL BEHAV. 671 (1972) (finding that the depth at which memory is "encoded," which is correlated with the amount of effort exerted, affects the durability of the memory).

(6.) See generally ENDEL TULVING, ELEMENTS OF EPISODIC MEMORY (1983) (finding that memory encoding is environment-specific, causing retrieval difficulties when the environment in which the subject is asked to recall the memory differs from the environment in which the memory was encoded).

(7.) See generally DANIEL KAHNEMAN, ATTENTION AND EFFORT (1973) (modeling attentive mental processing as a problem of scarce resource allocation); HAROLD E. PASHLER, THE PSYCHOLOGY OF ATTENTION (1998) (reviewing research on attention, including recent contributions). For a sampling of ongoing research on attention, see Barrett et al., supra note 2 (investigating individual differences in the ability to control attention).

(8.) See, e.g., Alan Hobbs & Ann Williamson, Associations Between Errors and Contributing Factors in Aircraft Maintenance, 45 HUM. FACTORS 186 (2003) (demonstrating a correlation between fatigue and cognitive error in aircraft maintenance); Dennis H. Holding, Fatigue, in STRESS AND FATIGUE IN HUMAN PERFORMANCE 145 (Robert Hockey ed., 1983) (reviewing the effects of fatigue on cognition).

(9.) I use the terms "evidentiary procedure" and "evidentiary process" to denote a hybrid field of law concerning the manner in which the legal system itself processes information. This field contains most of evidence and those portions of civil, criminal, and administrative procedure involving discovery, investigation, interrogation, and factfinding.

(10.) The following chronologically ordered list is but a small sample of this literature: PATRICK M. WALL, EYE-WITNESS IDENTIFICATION IN CRIMINAL CASES (1965) (discussing how, in identifying suspects in a police lineup, eyewitnesses are apt to have "memory, source confusion," mistaking familiarity with a suspect due to prior viewing of police photos for familiarity due to sighting in relation to the crime), cited with approval in United States v. Wade, 388 U.S. 218, 229, 230, 232, 234 (1967) and Simmons v. United States, 390 U.S. 377, 383, 386, 388 (1968): I. Daniel Stewart, Jr., Perception, Memory. and Hearsay: A Criticism of Present Law and the Proposed Federal Rules of Evidence, 1970 UTAH L. REV. 1 (1970) (discussing eyewitness fallibility generally); Muriel D. Lezak, Some Psychological Limitations on Witness Reliability, 20 WAYNE L. REV. 117 (1973) (same); Evan Brown, Kenneth Deffenbacher & William Sturgill, Memory for Faces and the Circumstances of. Encounter, 62 J. APPLIED PSYCHOL. 311 (1977) (discussing memory-source confusion in relation to eyewitnesses); ELIZABETH F. LOFTUS, EYEWITNESS TESTIMONY 20-109 (1st ed. 1979) (discussing eyewitness fallibility generally); Gary L. Wells, R.C.L. Lindsay & Tamara Ferguson, Accuracy, Confidence, and Juror Perceptions in Eyewitness Identification, 64 J. APPLIED PSYCHOL. 440 (1979) (identifying the "confidence-accuracy correlation" problem: the weak correlation between eyewitnesses' confidence in the accuracy of their memories and the actual accuracy of those memories); Stephan Landsman, Reforming Adversary Procedure: A Proposal Concerning the Psychology of Memory and the Testimony of Disinterested Witnesses, 45 U. PITT. L. REV. 547 (1984) (discussing various memory faults in relation to witness coaching): John S. Applegate, Witness Preparation, 68 TEX. L. REV. 277 (1989) (same); R.C.L. Lindsay, Gary L. Wells & Fergus O'Connor, Mock-Juror Belief of Accurate and Inaccurate Eyewitnesses, 13 LAW & HUM. BEHAV. 333 (1989) (presenting findings on the confidence-accuracy correlation problem); Steven I. Friedland, On Common Sense and the Evaluation of Witness Credibility, 40 CASE W. RES. L. REV. 165 (1990) (arguing that jurors are poor at assessing the accuracy of testimony); Russell Korobkin & Chris Guthrie, Psychological Barriers to Litigation Settlement: An Experimental Approach, 93 MICH. L. REV. 107 (1994) (arguing that cognitive limits hamper the settlement of disputes); BRIAN L. CUTLER & STEVEN D. PENROD, MISTAKEN IDENTIFICATION: THE EYEWITNESS, PSYCHOLOGY, AND THE LAW (1995) (discussing the fallibility of eyewitness identification generally); Jeffrey J. Rachlinski, A Positive Psychological Theory of Judging in Hindsight, 65 U. CHI. L. REV. 571 (1998) [hereinafter Rachlinski, Hindsight] (arguing that judicial factfinders are prone to "hindsight bias," defined infra note 15); Jeffrey J. Rachlinski, Heuristics and Biases in the Courts: Ignorance or Adaptation?, 79 OR. L. REV. 61 (2000) [hereinafter Rachlinski. Adaptation?] (arguing that factfinders are prone to "hindsight bias" as well as errors caused by use of the "representativeness heuristic," defined infra note 17); Deborah Davis & William C. Follette, Foibles of Witness Memory for Traumatic/High Profile Events, 66 J. AIR L. & COM. 1421 (2001) (surveying research on eyewitness fallibility); Chris Guthrie, Jeffrey J. Rachlinski & Andrew J. Wistrich, Inside the Judicial Mind, 86 CORNELL L. REV. 777 (2001) (discussing cognitive illusions in judges' decisionmaking); Bennett L. Gershman, Witness Coaching by Prosecutors, 23 CARDOZO L. REV. 829, 838-44 (2002) (discussing various memory faults in relation to witness coaching); D. Michael Risinger & Jeffrey L. Loop, Three Card Monte, Monty Hall, Modus Operandi. and "Offender Profiling": Some Lessons of Modern Cognitive Science for the Law of Evidence, 24 CARDOZO L. REV. 193, 194 (2002) ("[T]he last century has seen the accumulation of literally thousands of studies on the weaknesses of eyewitness testimony...."); AMINA MEMON, ALDERT VRIJ & RAY BULL, PSYCHOLOGY AND LAW: TRUTHFULNESS, ACCURACY AND CREDIBILITY 87-167 (2d ed. 2003) (surveying research on eyewitness fallibility); Erica Beecher-Monas, Heuristics, Biases, and the Importance of Gatekeeping, 2003 MICH. ST. L. REV. 987 (2003) (discussing heuristics and biases in jury decisiomnaking); Michael J. Saks & D. Michael Risinger, Baserates, The Presumption of Guilt, Admissibility Rulings, and Erroneous Convictions, 2003 MICH. ST. L. REV. 1051 (2003) (discussing decisionmakers' errors in revising their probabilistic assessments in light of new evidence--specifically, their failure to fully account for underlying population frequencies, or "baserates").

(11.) See, e.g., LOFTUS, supra note 10.

(12.) See, e.g., Lindsay et al., supra note 10; Wells et al., supra note 10.

(13.) See, e.g., Lindsay et al., supra note 10; Wells et al., supra note 10.

(14.) Importantly, specific cognitive illusions and biases are grounded in more general limits on cognitive capacity. Christine Jolls, Cass R. Sunstein & Richard Thaler, A Behavioral Approach to Law and Economics, 50 STAN. L. REV. 1471, 1477-78 (1998) ("We have limited computational skills and seriously flawed memories.... To deal with [this,] we use mental shortcuts and rules of thumb ... [which, in turn,] lead us to erroneous conclusions."); Amos Tversky & Daniel Kahneman, Judgment Under Uncertainty: Heuristics and Biases, 185 SCIENCE 1124, 1124-27 (1974) ("[P]eople rely on ... heuristic principles which reduce ... complex tasks ... to simpler judgmental operations. In general, these heuristics are quite useful, but sometimes they lead to severe and systematic errors.").

(15.) See Baruch Fischhoff, For Those Condemned to Study the Past: Heuristics and Biases in Hindsight, in JUDGMENT UNDER UNCERTAINTY: HEURISTICS AND BIASES 335, 341 (Daniel Kahneman, Paul Slovic & Amos Tversky eds., 1982) ("In hindsight, people consistently exaggerate what [they or others] could have ... anticipated in foresight.").

(16.) See. e.g., Rachlinski, Adaptation?, supra note 10. But see Chris W. Sanchirico, Finding Error, 2004 MICH ST. L. REV. 1189 (2004) (identifying rational twin of across-person hindsight bias).

(17.) See Rachlinski, Adaptation?, supra note 10, at 82 ("The representativeness heuristic refers to the reliance on the degree of apparent similarity between the features of the events to the features of the category in judging whether an event is a member of a particular category."); Amos Tversky & Daniel Kahneman, Judgments of and by Representativeness, in JUDGMENT UNDER UNCERTAINTY, supra note 15, at 84-87.

(18.) Rachlinski, Adaptation?, supra note 10, at 81-85; Saks & Risinger, supra note 10, at 1056-57. Strikingly versatile, the representativeness heuristic also causes factfinders to "make categorical predictions that are insensitive to the reliability of the evidence supporting their predictions, misunderstand what a random sequence of events looks like, and fail to appreciate common statistical phenomena like regression to the mean." Rachlinski, Adaptation?, supra note 10, at 85; see also Tversky & Kahneman, supra note 17, at 84-87, 98 (describing specific errors due to the representativeness heuristic, including a hypothetical example wherein the jury fails to understand that the event, "the defendant left the scene of the crime," must be more likely than the event, "the defendant left the scene of the crime for fear of being accused of murder").

(19.) See, e.g. (in chronological order), Cass R. Sunstein, Behavioral Analysis of Law, 64 U. CHI. L. REV. 1175, 1194 (1997) ("Economic analysis of law has proceeded on the basis of inaccurate understandings of decision and choice."); Jolls et al., supra note 14 (similar); Jon D. Hanson & Douglas A. Kysar, Taking Behavioralism Seriously: The Problem of Market Manipulation, 74 N.Y.U. L. REV. 630, 630 (1999) ("[S]cholars have been well justified in incorporating the behavioralist account of human behavior into law and economics. Nevertheless ... those scholars simultaneously have failed to take the findings of behavioral research to their logical conclusion."); Russell B. Korobkin & Thomas S. Ulen, Law and Behavioral Science: Removing the Rationality Assumption from Law and Economics, 88 CAL. L. REV. 1051, 1053 (2000) ("As law and economics turns forty years old, its continued vitality is threatened by its unrealistic core behavioral assumption: that people subject to the law act rationally.").

(20.) Hanson & Kysar, supra note 19, at 727 ("[I]ndividuals['] ... departure from models of scientific and statistical reasoning ... creates an opportunity for manufacturers to ... exploit consumer perceptions.").

(21.) See R.H. Coase, The Problem of Social Cost, 3 J.L. & ECON. 1, 4 (1960).

(22.) Jolls et at., supra note 14, at 1497-1501 (arguing that the "endowment effect" may prevent bargaining, though questioning the effect on efficiency if the endowment effect changes preferences).

(23.) See, e.g., United States v. Wade, 388 U.S. 218, 228-37 (1967) (citing WALL, supra note I0, in finding a Sixth Amendment right to the presence of counsel at postindictment lineups); RONALD P. FISHER & R. EDWARD GEISELMAN, MEMORY-ENHANCING TECHNIQUES FOR INVESTIGATIVE INTERVIEWING: THE COGNITIVE INTERVIEW (1992) (applying "cognitive interview" techniques to improve the questioning of witnesses).

(24.) This Article claims instrumental value only for limits on information-processing capacity, not for limits on information. Correspondingly, the literature to which the Article responds rests its negative prognosis not on a lack of information, but on misuse of what information exists.

(25.) The instrumental value hypothesis should be distinguished from two orthogonal hypotheses. The first is that the law has in some cases "adapted" to cognitive limitations. See, e.g., 1A JOHN HENRY WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW [section] 58.2, at 1212-34 (Peter Tillers ed., 4th ed. 1983) (justifying restrictions on character evidence as an adaptation to jury overweighing); Rachlinski, Hindsight, supra note 10, at 575 ("The law has adapted well to the fallibility of human judgment."). The difference between the adaptation hypothesis and the instrumental value hypothesis is the difference between mitigation and exploitation. One is worse off for having had to adapt; one is better off for having been able to exploit. The second hypothesis--whose difference is self-evident--is that the prevalence and impact of cognitive error have been overstated. See, e.g. (in chronological order), David C. Funder, Errors and Mistakes: Evaluating the Accuracy of Social Judgment, 101 PSYCHOL. BULL. 75, 84 (1987) (critiquing findings of error in social judgment); Gerd Gigerenzer, How to Make Cognitive Illusions Disappear: Beyond "Heuristics and Biases," 2 EUR. REV. SOC. PSYCHOL. 83 (1991) (finding errors overstated); Gerd Gigerenzer & Peter M. Todd, Fast and Frugal Heuristics: The Adaptive Toolbox, in SIMPLE HEURISTICS THAT MAKE US SMART 3, 21-22, 29-31 (Gerd Gigerenzer & Peter M. Todd eds., 1999) (similar); Ebbe B. Ebbesen, Some Thoughts About Generalizing the Role That Confidence Plays in the Accuracy of Eyewitness Memory (Nov. 3, 2000) (critiquing the experimental design of confidence-accuracy correlation studies), at; Gregory Mitchell, Taking Behavioralism Too Seriously? The Unwarranted Pessimism of the New Behavioral Analysis of Law, 43 WM. & MARY L. REV. 1907, 1945-95 (2002) (faulting cognitive errors literature for experimental design, statistical method, and ecological validity); People v. Legrand, 747 N.Y.S.2d 733 (Sup. Ct. 2002) (holding expert testimony on eyewitness fallibility not "generally accepted" under a Frye standard); Gregory Mitchell, Mapping Evidence Law, 2003 MICH. ST. L. REV. 1065 (2003) (critiquing applications of the errors literature to evidence law).

(26.) See Oliver W. Holmes, The Path of the Law, 10 HARV. L. REV. 457 (1897) (proposing that the law be analyzed from the perspective of the "bad man"); see also Charles R. Nesson, Incentives to Spoliate Evidence in Civil Litigation: The Need for Vigorous Judicial Action, 13 CARDOZO L. REV. 793, 795, 805 (1991) (advocating Holmes's "bad man" approach in the context of evidentiary process).

(27.) E.g., Margo v. Weiss, 213 F.3d 55 (2d Cir. 2000) (sanctioning plaintiffs for lying in their affidavit regarding the timing of their claim); Jones v. Clinton, 36 F. Supp. 2d 1118, 1121-25 (E.D. Ark. 1999) (finding that President Clinton lied under oath).

(28.) See, e.g., Superseding Indictment, United States v. Stewart, 323 F. Supp. 2d 606 (S.D.N.Y. 2004) (S1 03 Cr. 717 (MGC)), available at (charging, inter alia, that domestic icon Martha Stewart altered, but then restored, phone logs and that Stewart's broker, Peter Bacanovic, altered customer worksheets); Constance L. Hays & Leslie Eaton, Stewart Found Guilty of Lying in Sale of Stock, N.Y. TIMES, Mar. 6, 2004, at A1.

(29.) See, e.g., Indictment, United States v. Arthur Andersen, LLP, 2002 U.S. Dist. LEXIS 26870 (S.D. Tex. 2002) (No. H-02-121) (on file with author); Kurt Eichenwald, Andersen Misread Depths of the Government's Anger, N.Y. TIMES, Mar. 18, 2002, at A1 (describing Arthur Andersen's shredding of documents relating to Enron's special purpose entities).

(30.) See, e.g., MARY W. SHELLY, FRANKENSTEIN (1818); Sir Arthur Conan Doyle, The Final Problem, in 1 THE COMPLETE SHERLOCK HOLMES 469, 470-71 (1960) (depicting Sherlock Holmes as holding the following opinion of Professor Moriarty: "[A] man of ... excellent education, endowed by nature with a phenomenal mathematical faculty.... [But] a criminal strain ran in his blood, which, instead of being modified, was increased and rendered infinitely more dangerous by his extraordinary mental powers.").

(31.) A similar point might be made about other "bads," including coordination problems and uncertainty. Regarding the benefits of uncertainty, see Chris W. Sanchirico, Evidence Tampering, 53 DUKE L.J. 1215, 1305, 1306 (2004) ("[T]he clearer the parties' sense of ... which [evidence] will be decisive in future litigation, the more effectively they can target their destruction and fabrication efforts."). Regarding the benefits of coordination problems, see infra Part II.B.4 (concerning witness exclusion).

(32.) See, e.g., Donald A. Norman, Cognitive Artifacts, in DESIGNING INTERACTION 17, 17 (John M. Carroll ed., 1991) ("A cognitive artifact is an artificial device designed to maintain, display, or operate upon information in order to serve a representational function."). But see Edwin Hutchins, Cognitive Artifacts, in THE MIT ENCYCLOPEDIA OF THE COGNITIVE SCIENCES 126, 127 (Frank Keil & Robert Wilson eds., 1999) ("[There is] no widespread consensus on how to bound the category 'cognitive artifacts.'").

(33.) See, e.g., Chris W. Sanchirico, Character Evidence and the Object of Trial, 101 COLUM. L. REV. 1227 (2001) (arguing that character evidence rules are best explained and evaluated according to their effect on primary-activity incentives); Sanchirico, supra note 31 (arguing similarly for the law's regulation of perjury, obstruction, evidence destruction, and the like).

(34.) See supra note 25 (containing discussion of the adaptation hypothesis).

(35.) See, e.g., WEBSTER'S THIRD NEW INTERNATIONAL DICTIONARY OF THE ENGLISH LANGUAGE, UNABRIDGED 2362 (1993) [hereinafter WEBSTER'S THIRD NEW INTERNATIONAL DICTIONARY] (referring the reader, in etymological references for the word "testify," to the entry for "testament," which reads in part, "L[atin] testis" akin to Oscan trstus witnesses; both fr[om] a prehistoric Italic compound whose first and second constituents respectively are akin to L[atin] tres three and to L[atin] stare to stand; fr[om] the witness standing by as a third party in a litigation"); accord 17 OXFORD ENGLISH DICTIONARY, UNABRIDGED 829 (2d ed. 1989).


(37.) See, e.g., id. ("ME, vessel in which metals were assayed ... fr[om] L[atin] testum earthen vessel; akin to L[atin] testa piece of burned clay, earthen pot, shell.").

(38.) The testum appears to have been used to separate metal composites as much as to test composition.

(39.) This at least is what appears from the common etymological references. See, e.g., WEBSTER'S THIRD NEW INTERNATIONAL DICTIONARY, supra note 35, at 2362; accord 17 OXFORD ENGLISH DICTIONARY, supra note 35, at 829. In fact, it seems doubtful that "testify" is completely unrelated to "test," but I will leave this issue aside.

(40.) Davis & Follette, supra note 10, at 1435 n.19 (providing an extensive up-to-date survey of research on eyewitness fallibility, citing hundreds of articles and scores of results, and yet stating in its introduction that "[t]hroughout the article, the discussion will be confined to sources of unintentional errors"); Landsman, supra note 10, at 549 n.8 ("[Disinterested] eyewitnesses have been the subject of the most extensive judicial comment and psychological study. The reason for our preoccupation with eyewitnesses is not entirely clear.... For whatever reason, the literature on eyewitnesses is by far the most abundant, detailed and sophisticated.").

(41.) See, e.g., Applegate. supra note 10 (discussing memory shortcomings in relation to witness coaching); Gershman, supra note 10, at 838-44 (same); Landsman, supra note 10 (same). But see Charles Silver, Preliminary Thoughts on the Economics of Witness Preparation, 30 TEX. TECH. L. REV. 1383 (1999).

(42.) Cf. Roger C. Park, Adversarial Influences on the Interrogation of Trial Witnesses, in ADVERSARIAL VERSUS INQUISITORIAL JUSTICE 131, 153-58 (Peter J. van Koppen & Steven D. Penrod eds., 2003) (using the word "test" in a more literal sense).

(43.) To be sure, this was not the case prior to nineteenth-century reforms. George Fisher, The Jury's Rise as Lie Detector, 107 YALE L.J. 575 (1997), provides historical background for much of the analysis in this Part, including the present discussion of interested witnesses and subsequent treatments of the oath, the ordeal, and perjury.

(44.) U.S. CONST. amend. V; United States v. Echeles, 352 F.2d 892, 897 (7th Cir. 1965).

(45.) HARRY KALVEN, JR. & HANS ZEISEL, THE AMERICAN JURY 143-45 (1966), presents survey evidence--albeit dated--indicating that "the defendant testifies in 82 per cent of all cases." Id. at 144. More recent systematic empirical evidence on the related issue of how often suspects waive their Miranda rights--including the right not to provide the state with evidence of a testimonial nature during custodial interrogation--points in the same direction. See, e.g., Paul G. Cassell & Bret S. Hayman, Police Interrogation in the 1990s." An Empirical Study of the Effects of Miranda, 43 U.C.L.A.L. REV. 839 (1996).

(46.) In re Raiford, 695 F.2d 521, 523 (11th Cir. 1983) ("The use of a criminal conviction as conclusive of an issue in subsequent civil litigation, though not universally accepted, is well established today.").

(47.) See, e.g., 140 CONG. REC. H8991-92 (daily ed. Aug. 21, 1994) (statement of Rep. Molinari) (depicting sexual assault cases as "unresolvable swearing matches"); David Karp, Evidence of Propensity and Probability in Sex Offense Cases and Other Cases, 70 CHL-KENT L. REV. 15, 20 (1994) (pointing to "the typically secretive nature of [sex offense] crimes, and resulting lack of neutral witnesses in most cases").

(48.) See, e.g., 140 CONG. REC. H8990-92 (daily ed. Aug. 21, 1994) (statement of Rep. Molinari) (providing principal House sponsor's justification for enactment of Federal Rules of Evidence 413-415, which allow evidence of similar crimes in sexual assault and child molestation cases).

(49.) See, e.g., Murphy v. Bonano, 663 A.2d 505, 506-08 (D.C. 1995) (similar facts).

(50.) See, e.g., Banker's Trust Co. v. Publicker Indus., Inc., 641 F.2d 1361, 1363 (2d Cir. 1981) (similar facts).

(51.) See, e.g., Steffan v. Cheney, 920 F.2d 74 (D.C. Cir. 1990) (similar facts); Capellupo v. FMC Corp., 126 F.R.D. 545, 546 (D. Minn. 1989) (similar facts).

(52.) See, e.g., JOEL BRINKLEY & STEVE LOHR, U.S. v. MICROSOFT 4-5 (2001) (similar facts).

(53.) See, e.g., United States v. Henry, 325 F.3d 93, 104 (2d Cir. 2001) (similar facts).

(54.) See, e.g., Davis & Follette, supra note 10, at 1441-42 ("[It is] uncontested among psychological researchers that ... memory is a function of the amount of attention.... [A]nything about the event ... or the state of those witnessing it that reduces either the amount or the quality of attention ... will also impair memory.").

(55.) KAHNEMAN, supra note 7.

(56.) See, e.g., Davis & Follette, supra note 10, at 1449 ("Attention is naturally drawn to things relevant to one's own personal interests, goals, or current concerns....").

(57.) See, e.g., Degelos v. Fid. & Cas. Co., 313 F.2d 809 (5th Cir. 1963) (similar facts).

(58.) See generally BRINKLEY & LOHR, supra note 52.

(59.) See, e.g., United States v. Sheffield, 55 F.3d 341, 342 (8th Cir. 1995) (similar facts).

(60.) See, e.g., United States v. Rinke, 778 F.2d 581 (10th Cir. 1985) (similar facts).

(61.) See, e.g., Davis & Follette, supra note 10, at 1447-48, 1451; Landsman, supra note 10, at 549 n.8.

(62.) Elizabeth F. Loftus, Eyewitness Testimony: Psychological Research and Legal Thought, 3 CRIME & JUST. 105, 128-32 (1981).

(63.) Davis & Follette, supra note 10, at 1430 ("Jurors tend to believe that witness testimony (particularly that of uninterested witnesses) is accurate."); Landsman, supra note 10, at 549 ("[T]estimony of disinterested witnesses is one of the most persuasive sources of evidence in any lawsuit. Such testimony, most frequently eyewitness statements, is decisive in a large percentage of the cases in which it is offered.").

(64.) 7 Car. & Payne's Rep. 350 (Carmarthen Assizes 1836).

(65.) Id. at 350-51. The opinion focuses on the question whether opposing counsel could offer an additional witness to contradict the witness's denial of the relationship. The answer was yes.

(66.) 469 U.S. 45 (1984).

(67.) Id. at 47.

(68.) Id. at 47, 54-56.

(69.) U.S. CONST. amend. VI ("In all criminal prosecutions, the accused shall enjoy the right ... to be confronted with the witnesses against him...."): Olden v. Kentucky, 488 U.S. 227, 231 (1988) ("[E]xposure of a witness' motivation in testifying is a proper and important function of the constitutionally protected [Sixth Amendment] right of cross-examination."); Delaware v. Van Arsdall, 475 U.S. 673. 680 (1986) (similar): Davis v. Alaska, 415 U.S. 308,316-17 (1974) (similar).

(70.) I mean to define the preposition "before" broadly to include other statements by the individual that may be admissible for some purpose. See infra Parts II.B.2-3.

(71.) The analysis will draw on the notion of "costly signaling" as introduced into economics by A. MICHAEL SPENCE, MARKET SIGNALING: INFORMATIONAL TRANSFER 1N HIRING AND RELAYED SCREENING PROCESSES (1974), and as extended to "endogenous cost signaling" and used to model evidence production in Chris W. Sanchirico, Games, Information, and Evidence Production: With Application to English Legal History, 2 AM. L. & ECON. REV. 342, 348 (2000) [hereinafter Sanchirico, Games]; Chris W. Sanchirico, Relying on the Information of Interested--and Potentially Dishonest--Parties, 3 AM. L. & ECON. REV. 320, 329 (2001) [hereinafter Sanchirico, Relying]; Sanchirico, supra note 31; and Chris W. Sanchirico, Enforcement by Hearing: How the Civil Law Sets Incentives (1995) (Colum. Econ. Dept. Discussion Paper No. 95-9603, and first circulated version of Sanchirico, Relying and Sanchirico, Games) (on file with author) [hereinafter Sanchirico, Enforcement by Hearing].

(72.) FED. R. EVID. 603.

(73.) Two technical notes are in order here: (1) we measure the cost incrementally, relative to the cost of other possible answers, and (2) the reward here is measured in terms of upticks in the favorability of the "litigation lottery." The "litigation lottery" is the list of possible litigation outcomes, each paired with its probability.

(74.) This discussion implicates to some extent the game theory concept of "cheap talk." See, e.g., Joseph Farrell & Matthew Rabin, Cheap Talk, 10 J. ECON. PERSP. 103 (1996). A signal choice is said to be cheap talk when it "does not directly affect payoffs." Id. at 104. But this commonly stated definition is unhelpfully terse. For example, it does not account for the fact that there are two dimensions to contend with: (1) the underlying truth (or "type"), and (2) the signal. Referencing the discussion in the text, if "yes" were more expensive than "no" for truthful witnesses and equally so for untruthful witnesses, this would presumably not be "cheap talk." Yet the two types of witnesses would make the same choices and receive the same payoffs. Conversely, if answering "yes" cost the same as "no" for the truth-teller, and the same equality held for the liar, this would presumably satisfy the definition of "cheap talk." Yet such a signal would separate the types by payoffs, and so be helpful in setting primary-activity incentives, if the equal cost across signals for the truth teller were significantly lower than the equal cost across signals for the liar. Given this and other sources of imprecision, the current discussion eschews the term "cheap talk."

(75.) SPENCE, supra note 71.

(76.) The same holds when the witness and the proponent are not one and the same to the extent that the proponent bears some portion of the signal costs of the witness, as where the proponent and the witness interact in anything other than a one-shot relationship.

(77.) Sanchirico, Relying, supra note 71, at 344-45 (Proposition 1).

(78.) Notice here that primary-activity incentive setting does not require truth finding. The differential cost of lying acts like a tax on primary-activity choices that must be lied about to avoid punishment. See id. at 345 (Proposition 2).

(79.) I will be caricaturizing this process to make a more general point about testimony. For a more nuanced view, see, for example, Mirjan R. Damaska, Rational and Irrational Proof Revisited, 5 CARDOZO J. INT'L & COMP. L. 25 (1997); and Fisher, supra note 43.

(80.) "Formula Liturgicae" in Use at Ordeals (n.d.), in ERNEST F. HENDERSON, SELECT HISTORICAL DOCUMENTS OF THE MIDDLE AGES 314, 315 (1912).

(81.) Id.

(82.) See, e.g., ALDERT VRIJ, DETECTING LIES AND DECEIT: THE PSYCHOLOGY OF LYING AND THE IMPLICATIONS FOR PROFESSIONAL PRACTICE xiii (2000) (using the term "non-verbal behavior" as I use "demeanor").

(83.) Bella M. DePaulo, James J. Lindsay, Brian E. Malone, Laura Muhlenbruck, Kelly Charlton & Harris Cooper. Cues to Deception, 129 PSYCH. BULL. 74 app. A at 115-17 (2003) (listing, inter alia, over one hundred experimentally studied demeanor cues); see also VRIJ, supra note 82, at 32-33.

(84.) See the sources cited in Olin Guy Wellborn III, Demeanor, 76 CORNELL L. REV. 1075, 1076-78 (1991).

(85.) See id. (reviewing experimental results on the inefficacy of demeanor); see also Roger C. Park, Empirical Evaluation of the Hearsay Rule, in ESSAYS FOR COLIN TAPPER 91, 91-93 (Peter Mirfield & Roger Smith eds., 2003). But see James P. Timony, Demeanor Credibility, 49 CATH. U. L. REV. 903, 916-22 (2000) (defending demeanor evidence).

(86.) See, e.g., M. Zuckerman, B.M. DePaulo & R. Rosenthal, Verbal and Nonverbal Communication of Deception, 14 ADVANCES IN EXPERIMENTAL SOC. PSYCHOL. 1, 9-10 (1981) (discussing this hypothesis in the literature).

(87.) See sources on costly signaling cited supra note 71. The lie detection literature is starting to embrace the logic of costly signaling. See, e.g., VRIJ, supra note 82, at 38 ("[B]ehaviours which are easy to control can never be reliable indicators of deception.").

(88.) DePaulo et al., supra note 83, at 93 ("The 32 independent estimates of eye contact produced a combined effect that was almost exactly zero.").

(89.) A meta-analysis quantitatively combines the results of prior studies that meet express criteria. See R. Rosenthal & M.R. DiMatteo, Meta-analysis: Recent Developments in Quantitative Methods for Literature Reviews, 52 ANN. REV. PSYCHOL. 59 (2001) (providing a primer on meta-analysis).

(90.) DePaulo et al., supra note 83, app. A at 115-17.

(91.) Id. (listing the cues considered).

(92.) Id. pt. VI.A (listing criteria for inclusion); id. at 85 tbl.1 (listing the prior studies included).

(93.) DePaulo et al., supra note 83, measure "effect size" with Cohen's d. See JACOB COHEN, STATISTICAL POWER ANALYSIS FOR THE BEHAVIORAL SCIENCES 20-24 (2d ed. 1988). Roughly, d is the proportion of deviation from the mean explained by the difference between lying versus truth telling, as opposed to "noise" factors such as individual and circumstantial differences.

In fact, d is of questionable utility. The size of d says little about the size of a cue's likelihood ratio, the true measure of the cue's informativeness. The cue's likelihood ratio is the ratio of the probability that a liar exhibits the cue to the probability that a truth teller exhibits the cue. According to Bayes's rule, on seeing the cue, one multiplies the prior odds that the subject is lying (the ratio of the chance that she's lying to the chance that she's not, prior to revelation of whether the cue is present) by the likelihood ratio in order to obtain the posterior odds that the subject is lying. Thus, the extent to which a cue should affect our assessment of sincerity is precisely measured by the likelihood ratio, and not d.

(94.) Cohen, upon whom DePaulo et al., supra note 83, rely, see supra note 93, proposes that a d between -.20 and +.20 is a "small effect size." COHEN, supra note 93, at 25. To intuitively calibrate "small effect size," Cohen points out that in judging whether a human female is fifteen or sixteen years old, height has a small effect size. But see the discussion of d's limitations at note 93 supra.

(95.) DePaulo et al., supra note 83, at 95 tbl.8. Over the last decade, several relatively obscure demeanor cues have been found to exhibit larger effect sizes. Paul Ekman's work has focused on "microfacial expressions," small movements of the facial muscles detectable only for a fraction of a second. PAUL EKMAN, TELLING LIES: CLUES TO DECEIT IN THE MARKETPLACE, POLITICS, AND MARRIAGE 123-61 (2001). Such cues are specifically not detectable by lay observers, although lay observers, it is asserted, can be trained to notice them. Additionally, Zuckerman et al., supra note 86, ask whether the masking of demeanor cues might itself be detectable, as when liars appear unnaturally "stiff." Game theoretic reasoning suggests that findings of modest success for these obscure cues ought to be regarded with skepticism, precisely because they are obscure. A large effect size says little if experimental subjects did not realize that the cue was being watched.

(96.) DePaulo et al., supra note 83, at 100.

(97.) One could always claim that, even though every witness will experience stress at trial, the guilty or insincere will be even more stressed and, consequently, stress cues will still have utility. Id. at 96, 100, 102. Whether or not this proposition holds, existing findings provide no support for it.

(98.) VRIJ, supra note 82, at 170. Sometimes brain electrical behavior is monitored as well. Id.

(99.) This is clearest in relation to the Control Question Test, one variant of polygraph testing. Explanations of the logic behind the Control Question Test are not entirely coherent. EKMAN, supra note 95, at 200. The following note contains the author's own attempt to make sense of the test mechanic.

The Control Question Test pairs "relevant questions" with "control questions." If the issue is whether the subject assaulted the victim, a relevant question might be, "Did you assault the victim?," and the control question might be, "Have you ever struck anyone in anger?" A control question is designed to be one that the subject believes is nearly as pertinent to her interests in the present suit as the corresponding relevant question--as if similar past transgressions will be used against her in this case. Moreover, the control question should be one that nearly all subjects, innocent or guilty, would answer insincerely.

Consider how this plays out--first for the innocent subject, then for the guilty. In falsely answering the control question, the innocent subject experiences two sources of stress: from the pertinence of the question and from lying. In truthfully answering the relevant question, on the other hand, the innocent subject experiences only "pertinence stress." Plausibly, pertinence stress from the relevant question is greater than for the control question. Lying stress, on the other hand, is greater for the control question, on which the innocent subject lies, than for the relevant question, on which she tells the truth. The assumption is that the difference in lying stress dominates so that the innocent subject exhibits more total stress in response to control questions. The guilty subject, on the other hand, lies on both questions. For the guilty subject, therefore, the greater pertinence stress of the relevant question is decisive, and total stress is greater for the relevant question than for the control question. Thus, guilty subjects will exhibit more stress on the relevant question, while innocent subjects will exhibit more stress on the control question. Id.

Obviously, this truth-finding mechanic is a bit fragile, as has been pointed out repeatedly in the literature. To take a somewhat novel critical angle, consider that it must simultaneously obtain that: (1) the past transgression forming the subject of the control question is so common that the examiner can be fairly certain that any innocent subject has committed it, despite her denial; and (2) the subject believes that this past transgression--in fact so common that any innocent subject who walked into the room would have committed it--is now going to be regarded by the examiner as informative of whether the subject committed the particular crime in question. The difficulty of finding this sort of "magic bullet" question produces a serious correlation problem. Perhaps the transgression is so commonplace that the innocent subject doubts its pertinence to the present case. Or, perhaps the transgression is not common enough, and when the innocent witness answers "no," she is not really lying. The shared result of both of these complementary sources of error--i.e., that innocent subjects are insufficiently stressed by the control question relative to the relevant question--is consistent with the data. Both laboratory studies, VRIJ, supra note 82, at 197, and field studies, id. at 199, show that the Control Question Test has a roughly twenty percent chance of finding the innocent guilty.

(100.) Similarly, for the Control Question Test, described in note 99, supra, the guilty subject can learn to identify control questions and produce artificial stress in answering them.

(101.) VRIJ, supra note 82, at 203 (listing studies showing that thirty minutes of training in countermeasures can be effective and that such measures can be difficult to detect). If few individuals have been taught thus far, this may merely reflect the test's relative lack of employment.

(102.) See, e.g., 18 U.S.C. [subsection] 1621-1623 (2000) (federal perjury); 18 U.S.C. [subsection] 1503, 1512(c) (2000) (federal obstruction of justice); U.S. SENTENCING COMMISSION GUIDELINES MANUAL [subsection] 2J1.2 ("offense level" for federal obstruction of justice for purposes of sentencing), 2J1.3 ("offense level" for federal perjury for purposes of sentencing), 5E1.2 (2004) (fines for individuals based on "offense level" and other factors), ch. 5, pt. A (months of imprisonment based on "offense level" and criminal history). For a review of these laws, see Sanchirico, supra note 31, pt. II.

(103.) Sanchirico, Relying, supra note 71, at 330-31.

(104.) There may be other reasons to exclude these forms of evidence, including lack of scientific reliability, the risk that the jury will overweigh the evidence, and the fact that such evidence "invades the province of the jury." See, e.g., JACK B. WEINSTEIN & MARGARET A. BERGER, WEINSTEIN'S FEDERAL EVIDENCE [section] 702.06[1][b] (reviewing reasons why polygraph evidence may be excluded).

(105.) See, e.g., FED. R. EVID. 401 (defining relevance in this way).

(106.) VRIJ, supra note 82, at 109.

(107.) Id. at 210, 212, 215; Sanchirico, Games, supra note 71, at 348; Sanchirico, Relying, supra note 71, at 329; Sanchirico, Enforcement by Hearing, supra note 71.

(108.) DePaulo et al., supra note 83, at 91.

(109.) Id. at 96.

(110.) Id. at 92.

(111.) In certain circumstances, the prior statement may also be used to prove the truth of the matter asserted. See infra note 171.

(112.) The general rule, however, is that "extrinsic evidence" may not be used to contradict the witness if the matter being contradicted is "collateral." GRAHAM C. LILLY, AN INTRODUCTION TO THE LAW OF EVIDENCE [section] 8.5, at 415 (3d ed. 1996). As noted, "extrinsic" evidence is evidence outside the four corners of the witness's own testimony (including cross-examination), such as the testimony of other witnesses or documents and things. A matter is "collateral" if it could not he offered for any purpose other than impeachment or, alternatively, if it is not material to a claim or defense. Id. [section] 8.6, at 424.

(113.) See supra note 112 (defining "extrinsic evidence").

(114.) Compare, for instance, Federal Rule of Evidence 608 (which allows extrinsic evidence of untruthful character in the form of opinion and reputation witnesses, but not in the form of specific instances) and 609 (which allows extrinsic evidence of untruthful character in the form of evidence of a criminal conviction) with Federal Rule of Evidence 613 (which allows extrinsic evidence of a prior inconsistent statement under special procedural conditions) and the traditional rule, still followed in many circuits despite Federal Rule of Evidence 613, which disallows extrinsic evidence of a prior inconsistent statement if the matter stated is "collateral" (as defined in note 112, supra). WEINSTEIN & BERGER, supra note 104, [section] 607.06.

(115.) See Sanchirico, supra note 33, at 1264-77.

(116.) H. Richard Uviller, Credence, Character, and the Rules of Evidence: Seeing Through the Liar's Tale, 42 DUKE L.J. 776, 816-17, 827 (1993).

(117.) See, e.g., CHRISTOPHER B. MUELLER & LAIRD C. KIRKPATRICK, EVIDENCE UNDER mE RULES 663 (4th ed. 2000) (stating that impeachment by contradiction "goes on every day"); WEINSTEIN &, BERGER, supra note 104, [section] 607.06[1] (stating that impeachment by contradiction is "a well-recognized technique").

(118.) JEROME FRANK, COURTS ON TRIAL: MYTH AND REALITY IN AMERICAN JUSTICE 86 (1950) ("[The witness] often detects what the lawyer hopes to prove at the trial. If the witness desires to have the lawyer's client win the case, he will often, unconsciously, mold his story accordingly. Telling and re-telling it to the lawyer, he will honestly believe that his story, as he narrates it in court, is true, although it importantly deviates from what he originally believed.") (emphasis added).

(119.) Landsman, supra note 10, at 549 n.8.

(120.) Id. at 555 ("The lawyer's influence will frequently be amplified because she is perceived by the witness as a person of high status....").

(121.) Gershman, supra note 10, at 829-30.

(122.) Id.

(123.) Applegate, supra note 10, at 279. But see Silver, supra note 41, at 1387 (arguing that witness preparation is infrequent).

(124.) See ALFRED N. WHITEHEAD & BERTRAND RUSSELL, PRINCIPIA MATHEMATICA (1927) (providing axiomatic system from which much of mathematics may be derived), summarized in JAMES DUGUNDJI, TOPOLOGY 17-21 (1970). But see Kurt Godel, Uber Formal Unentscheidbare Satze der Principia Mathematica und Verwandter Systeme I, in 38 MONATSHEFTE FUR MATH. U. PHYSIK 173 (1931) (proving that within a large class of axiomatic systems, any system that is "consistent," in the sense that no proposition is simultaneously true and false, is "incomplete," in the sense that the truth or falsity of some propositions that may be stated within the system cannot be determined).

(125.) The liar's tale is always in a sense "externally inconsistent." This is what makes it a lie. The issue here is whether the contradicted facts are ascertainable by the opposition and, ultimately, the factfinder. Note also that the distinction between internal and external inconsistency is not always sharp: the suspect whose alibi places her simultaneously in Seattle and Los Angeles contradicts both herself and the external fact that Seattle and Los Angeles have no geographical intersection.

(126.) See supra pp. 317-18.

(127.) Perjury, obstruction, contempt, and the subpoena power probably do little to inspire witnesses to be forthcoming with contradictable detail. The classic "I don't recall" or "I don't know" are practical safe havens under these strictures.

(128.) BRINKLEY & LOHR, supra note 52, at 14.

(129.) See generally George Mandler, Hypermnesia, Incubation, and Mind Popping: On Remembering Without Really Trying, in 15 ATTENTION & PERFORMANCE 1 (1994) (reviewing literature on hypermnesia and related phenomena).

(130.) Fanelli v. United States Gypsum Co., 141 F.2d 216, 217 (2d Cir. 1944) (Frank, J.) ("Common experience, the work of Proust and other keenly observant literary men, and recondite psychological research, all teach us that memory of things long past can be accurately restored in all sorts of ways."); Baker v. State, 371 A.2d 699, 705 n.11 (Md. Ct. Spec. App. 1977) ("Marcel Proust, in his monumental epic In Remembrance of Things Past, sat, as a middle-aged man, sipping a cup of lime-flavored tea and eating a madeleine, a small French pastry. Through both media, two long-forgotten tastes from childhood were reawakened. By association, long forgotten memories from the same period of childhood came welling and surging back. Once those floodgates of recall were opened, seven volumes followed.").

(131.) See, e.g., 1 MCCORMICK ON EVIDENCE [section] 9, at 28 (John William Strong ed., 4th ed. 1992) ("It is abundantly clear from everyday observation that the latent memory of an experience may be revived by an image seen, or a statement read or heard.... The recall of any part of a past experience tends to bring with it the other parts that were in the same field of awareness.... The effect of a reminder, encountered in reading a newspaper or in the conversation of a friend, which gives us the sensation of recognizing as familiar some happening which we had forgotten, and prompts our memory to bring back associated experience, is a frequently encountered process.").

(132.) See generally Mandler, supra note 129 (reviewing literature on hypermnesia, incubation, and mind popping); id. at 3-4 ("In all of these cases, access to some mental content is increased with little deliberative effort, such as a delay or the mere request for another recall; the individual is not actually trying to improve on the performance.").

(133.) Id. at 5, 8.

(134.) Id. at 8.

(135.) See generally Daniel L. Schacter, Kenneth A. Norman & Wilma Koutstaal, The Cognitive Neuroscience of Constructive Memory, 49 ANN. REV. PSCYHOL. 289 (1998) (reviewing recent research on constructive processes that distort memory); id. at 290 ("Contemporary cognitive psychologists have been especially concerned with constructive aspects of memory....").

(136.) Part III.A. 1 infra discusses the extent to which current law fosters surprise.

(137.) Imagine conservatively, for example, that each answer has three possible follow-up questions, and each question has three possible answers. Then, even if the witnesses were certain that questioning would not go beyond three rounds, there would be [3.sup.6] = 729 potential sequences of questions and answers to consider and 3 + 27 + 243 = 273 potential questions to which answers must be devised.

(138.) See Fernand Gobet, Chess, Psychology of, in MIT ENCYCLOPEDIA, supra note 32, at 113.

(139.) In the federal system, Federal Rule of Evidence 612 implicitly allows a witness to consult written records to refresh memory while testifying. Case law clarifies that such a writing may be consulted only for the purpose of refreshing an exhausted memory and that refreshed memory, and not the medium, must be the true source of further testimony. United States v. Weller, 238 F.3d 1215, 1221 (10th Cir. 2001) (noting that the "'court has the discretion to withhold any writing from a witness where the judge believes that the document will be a source of direct testimony rather than the key to refreshing the witness' independent recollection"); Thompson v. United States, 342 F.2d 137, 139 (5th Cir. 1965) (noting requirements that witness's memory be exhausted and that writing refresh it); NLRB v. Fed. Dairy Co., 297 F.2d 487, 488-89 (1st Cir. 1962) ("The witness should first testify ... that the paper does in fact have that effect [of refreshing memory and].... that his recollection is exhausted. Prerequisites i[n] prompting a witness, such as exhaustion of memory, are so axiomatic that they are rarely referred to except in passing."). See also the discussion of present recollection refreshed at notes 130, 131, and 148.

Federal Rule of Evidence 611(c) stipulates that leading questions--questions that suggest their answer--"should not be used on the direct examination of a witness" (though there are exceptions). Direct examination is consequently an exercise in recall, not merely recognition. Thus, instead of asking the passenger witness, "And then you saw the defendant's car weave into oncoming traffic, is that right?," the plaintiff's lawyers must ask something on the order of, "And then what happened?"

(140.) Fed. Dairy Co., 297 F.2d at 488 n.3 ("'Since the Narrative ... should represent actual Recollection.... it becomes necessary to forbid the use of various artificial written aids capable of misuse so as to put into the witness's mouth a story which is in effect fictitious and corresponds to no actual Recollection.'") (quoting 3 JOHN HENRY WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW [section] 758, at 125 (3d ed. 1940)).

(141.) This explanation is compelling in certain settings--such as where a child is being interviewed for the first time about a possible incident of molestation. Despite the fact that findings from the special setting of child witnesses are not readily transferable to most of the evidentiary process, these results tend to get broad play in the literature on cognition and evidentiary process. See. e.g., VRIJ. supra note 82, at 113-15 (discussing the troth-finding technique of "Statement Validity Assessment," initially designed for child witnesses).

(142.) See supra pp. 317-18.

(143.) BADDELEY, supra note 2, at 64-72.

(144.) Murray Glanzer & Anita R. Cunitz, Two Storage Mechanisms in Free Recall, 5 J. VERBAL LEARNING & VERBAL BEHAV. 351, 351 (1966).

(145.) See, e.g., Craik & Lockhart, supra note 5, at 671 (noting that depth of analysis determines durability).

(146.) See, e.g., Glanzer & Cunitz, supra note 144, at 357-60 (noting that "recency" effect dissipates with a brief filled delay between hearing and recalling).

(147.) See supra note 139.

(148.) FED. R. EVID. 612 ("Writing Used to Refresh Memory.... [I]f a witness uses a writing to refresh memory for the purpose of testifying.., while testifying ... an adverse party is entitled to have the writing produced at the hearing, to inspect it, to cross-examine the witness thereon, and to introduce in evidence those portions which relate to the testimony of the witness."). Note that Federal Rule of Evidence 612 is specifically made subject to the Jencks Act, 18 U.S.C. [section] 3500 (2000), as codified in part and extended in part in Federal Rule of Criminal Procedure 26.2. See 1 MCCORMICK ON EVIDENCE [section] 97, at 392-96 (John W. Strong ed., 5th ed. 1999) [hereinafter MCCORMICK ON EVIDENCE (5TH ED.)].

(149.) Present recollection refreshed has received little scholarly attention. For an exception, see Thomas M. Tomlinson, Note, Pattern-Based Memory and the Writing Used to Refresh, 73 TEX. L. REV. 1461, 1478 (1995) ("[There are] two ways that a writing used to refresh skews testimony: It makes a witness appear more confident, and it suggests testimony to the witness.").

(150.) See supra Part II.A.3.

(151.) FED. R. EVID. 612 advisory committee's note.

(152.) FED. R. EVID. 612.

(153.) The "interests of justice" proviso, however vague, was added by Congress and did not appear in the advisory committee's proposed rule. WEINSTEIN & BERGER, supra note 104, [section] 612 App.01..

(154.) 93 F.R.D. 138 (D. Del. 1982).

(155.) Id. at 144. Federal Rule of Evidence 612, like most Federal Rules of Evidence, applies to depositions taken under the Federal Rules of Civil Procedure. FED. R. CIV. P. 30(c); WEINSTEIN & BERGER, supra note 104, [section] 612.02[5].

(156.) Julian, 93 F.R.D. at 145.

(157.) Id. at 144-45; see also Berkey Photo, Inc. v. Eastman Kodak Co., 74 F.R.D. 613 (S.D.N.Y. 1977) (holding generally that if an attorney shows work product to a lay or expert witness prior to a deposition, then the work product becomes discoverable).

(158.) Julian, 93 F.R.D. at 144. Not all courts have taken this "strict waiver" approach. See WEINSTEIN & BERGER, supra note 104, [section] 612.05, for a discussion of alternative approaches, including the "'balancing approach" adopted by some courts.

(159.) WEINSTEIN & BERGER, supra note 104, [section] 612.06[2] ("When a witness has refreshed his or her recollection with privileged materials before testifying, most courts ... appear to hold that use of the material constitutes waiver of the privilege. Likewise, when a witness consults a writing embodying his or her own communication to counsel, and the testimony discloses a significant part of the communication, most courts find that the attorney-client privilege has been waived.").

(160.) See generally id. [section] 612.04[2][b][ii] (discussing other avenues for obtaining preparation materials when the witness is an expert).

(161.) FED. R. EVID. 705.

(162.) FED. R. CIV. P. 26(a)(2)(B).

(163.) See, e.g., Banker's Trust Co. v. Publicker Indus., Inc., 641 F.2d 1361 (2d Cir. 1981) (concerning action by investment bank to recover fee, wherein vice president of bank testified using a chronology of events, which the other side inspected, asked her about, but then declined to introduce into evidence).

(164.) WEINSTEIN & BERGER, supra note 104, [section] 612.02[3].

(165.) Ford v. Philips Elecs. Instruments Co., 82 F.R.D. 359, 361 (E.D. Pa. 1979) (establishing guidelines for the questioning of a nonexpert deponent regarding preparatory conversations with counsel: inter alia, that "[s]uch inquiry may not ... include questions that tend to elicit the specific questions posed to the witness..., the general line of inquiry pursued..., the facts to which.., counsel appeared to attach significance, or any other matter that reveals ... counsel's mental impressions"). If the deponent is counsel's client, such conversations will also be protected by attorney-client privilege. However, courts have held that conversations between counsel and an expert witness are not protected. See, e.g., Intermedics, Inc. v. Ventritex, Inc., 139 F.R.D. 384, 387 (N.D. Cal. 1991) ("[A]bsent an extraordinary showing of unfairness that goes well beyond the interests generally protected by the work product doctrine, written and oral communications from a lawyer to an expert that are related to matters about which the expert will offer testimony are discoverable, even when those communications otherwise would be deemed opinion work product.") (emphasis added).

(166.) Jos. Schlitz Brewing Co. v. Muller & Phipps (Haw.), Ltd., 85 F.R.D. 118, 120 n.2 (W.D. Mo. 1980) ("Adoption of a waiver theory seems dubious as a matter of policy, moreover, in that it would encourage the sophisticated prospective witness to avoid refreshing recollections prior to testifying, or to use a 'coach' who has examined the documents, rather than the documents themselves."); Applegate, supra note 10, at 345.

(167.) Applegate, supra note 10, at 345 ("Oral description of documents is the worst possible type of witness preparation for the legal system to encourage."); see also Schlitz Brewing Co., 85 F.R.D. at 120 n.2 ("It seems likely that truthful and accurate testimony by a prospective witness will be assisted by reviewing files, and the Court believes the useful discipline of such examination probably outweighs the danger of fabrication and mistake caused by reliance on undisclosed writings.").

(168.) Steve Lohr, Antitrust Case Is Highlighting Role of Email, N.Y. TIMES, Nov. 2, 1998, at CI.

(169.) Id.

(170.) The impact may differ as a matter of degree, however. For example, in the case of prior inconsistent statements, the "other" witness is the same as the current witness and, therefore, the contradiction is less likely to be accidental or excusable.

(171.) indeed, in some cases a prior statement may itself be admitted for the truth of the matter asserted, as when, for example, a party's own statements are offered against her or a witness's prior inconsistent statement was made under oath in a judicial proceeding or deposition and the witness may now be cross-examined concerning the statement. See, e.g., FED. R. EVID. 801(d)(1)-(2).

(172.) The general point here, that preparation may extend beyond that undertaken on the eve of trial, also holds for contradiction by purely external facts, which are also, but to a much lesser extent, subject to manipulation before the definite prospect of litigation arises.

(173.) Sanchirico, supra note 31, pt. IV (emphasizing the importance for the regulation of evidence tampering of uncertainty about what evidence will be decisive in litigation).

(174.) But see infra note 175 (discussing other difficulties that would befall even the perfectly rational witness).

(175.) See supra Part I.D. Cognitive limitations are not the only source of difficulty here. Difficulties are also caused by uncertainty per se, which plagues the perfectly rational as well as the ordinarily irrational. There is only some chance that one's current statement will be decisively damaging in later litigation. And sometimes the risk is rationally taken. Sanchirico, supra note 31, pt. V. Similarly, some part of the prevalence of contradictable prior statements is due to unavoidable trade-offs between effectively proceeding in the primary-activity task and avoiding the generation of an evidentiary trail--again a difficulty plaguing all degrees of rationality. To proceed effectively, for example, one may need to coordinate with others. Coordinating necessitates communicating, and communicating effectively may require sending e-mails that are later damaging in litigation. See also supra note 31; infra note 183.

(176.) See, e.g., FED. R. EVID. 613 (laying out procedures for impeaching witnesses by their prior inconsistent statements): Harris v. New York, 401 U.S. 222 (1971) (allowing impeachment of accused by prior inconsistent statements that were barred from substantive use by Miranda v. Arizona, 384 U.S. 436 (1966)).

(177.) 513 U.S. 150 (1995).

(178.) Bolstering--or supporting the witness's credibility before it has been attacked--is generally disallowed. CHRISTOPHER B. MUELLER & LAIRD C. KIRKPATRICK, EVIDENCE [section] 6.49 (3d ed. 2003). Rehabilitation--supporting the witness's credibility after it has been attacked--is allowed, but several special restrictions have emerged regarding rehabilitation, specifically rehabilitation by prior consistent statement. In Tome, the Court held inadmissible a child's prior statements, consistent with current testimony, that her father had sexually abused her. The Court laid down the general rule that such prior consistent statements would be admissible only when they were made before the motive to fabricate had arisen. Tome concerned the substantive, rather than rehabilitative, use of prior consistent statements the statements were offered to prove the truth of the matter asserted, rather than to rebut a charge that the child was fabricating on the stand, no such charge having been leveled by the defendant. (The precise issue was whether the statements were exempted from the definition of hearsay under Federal Rule of Evidence 801(d)(1)(B).) But the pre-motive requirement may well be applied to prior consistent statements when used solely for a rehabilitation purpose. WEINSTEIN & BERGER, supra note 104, [section] 607.09[2][b] ("[The prior] trend towards flexibility [in rehabilitating witnesses by means of their prior consistent statements] was dealt a mortal blow in Tome v. United States....").

(179.) WEINSTEIN & BERGER, supra note 104, [section] 607.07[3][c].

(180.) 1 MCCORMICK ON EVIDENCE (5TH ED.), supra note 148, [section] 34, at 74-75.

(181.) FED. R. EVID. 615. Witnesses who have a statutory right to be present--for example, victims in certain cases--also cannot be excluded. Id.

(182.) Id.

(183.) Thus, another "bad"--the problem of coordination among individuals--becomes a "good" when applied to evidentiary process. See supra notes 31,175.

(184.) In addition to this general practice, specific rules requiring or more heavily weighing multiple witnesses do exist. See, e.g., U.S. CONST. art. Ill, [section] 3, el. 1 ("No Person shall be convicted of Treason unless on the Testimony of two Witnesses to the same overt Act, or on Confession in open Court."); 18 U.S.C. [section] 1621 (2000) (perjury statute that has been interpreted to require two witnesses to prove offense). On the survival of the common law two-witness rule in [section] 1621, see Rebecca Kislak & John J. Donoghue, Perjury,, 36 AM. CRIM. L. REV. 957, 972-73 (1999) and sources cited therein.

(185.) This phrase is most often used in connection with justifying the general prohibition against hearsay, see, e.g., FED. R. EVID. 801, and in giving content to the Confrontation Clause of the U.S. Constitution, see. e.g., White v. Illinois, 502 U.S. 346, 356 (1992).

(186.) See supra Part I.B.3.

(187.) 5 WIGMORE, supra note 25, [section] 1367, at 32 (quoted, for example, in White v. Illinois, 502 U.S. at 356).

(188.) Others have also been dissatisfied with Wigmore's assertion, though for somewhat different reasons. Several scholars focus on cross-examination's potential for "false positives," wherein truthful witnesses are discredited. Professors John Langbein and Mirjan Damaska, for example, both warn of the possibility that cross-examination will make the truthful witness look like a liar. Mirjan Damaska, Presentation of Evidence and Factfinding Precision, 123 U. PA. L. REV. 1083, 1094 (1975) ("Even with the best of intentions on the cross-examiner's part, reliable testimony may easily be made to look debatable, and clear information may become obfuscated."); John H. Langbein, The German Advantage in Civil Procedure, 52 U. CHI. L. REV. 823, 833-34 (1985) ("[B]ecause cross-examination allows so much latitude for bullying and other truth-defeating stratagems, it is frequently the source of fresh distortion when brought to bear against truthful testimony."). Wigmore himself also suggested this possibility but viewed the problem as soluble. 5 WIGMORE, supra note 25, [section] 1367, at 32 ("[Cross-examination] may make the truth appear like falsehood. But this abuse of its power is able to be remedied by proper control.").

It may be worth emphasizing that the bare fact that cross-examination exhibits false positives is in and of itself no indictment. Only perfect evidence is free from error, and perfect evidence does not exist. Furthermore, in judging the efficacy of cross-examination, one must consider not only false positives, but also true positives: the extent to which liars' are discredited. The fact that one must consider both false positives and true positives in judging cross-examination is another way of pointing out the importance of focusing on the differential, rather than either of its components in isolation. (Note that false negatives are the complement of true positives.)

Another recent position on cross-examination emphasizes the lack of empirical evidence. This position is well represented in the writings of Professor Roger C. Park. Park, supra note 42; Roger C. Park, Visions of Applying the Scientific Method to the Hearsay Rule, 2003 MICH. ST. L. REV. 1149 (2003). Park recognizes that the discrepancy between Wigmore's optimism and modern skepticism is an empirical matter. And he maintains neutrality given the dearth of empirical evidence for either side. Park, supra note 42, at 162 (reviewing a famous practice guide on cross-examination, as well as systematic empirical work, and concluding that "no one has come close to doing a definitive study [of cross-examination], and one could argue that no one has even made a good start"); see also Applegate, supra note 10, at 311 ("While the adversary system touts the effectiveness of cross-examination for revealing the truth, there is little empirical support for this conclusion."). Empirical evidence is certainly lacking and ultimately necessary. But what must first be supplied is a precise account of the hypotheses to be tested.


(190.) Id.

(191.) Steam engines had already been in use for several decades. But the engine was likely still at the edge of the technological consciousness, much like computer processing or nuclear power are today, despite the fact that they have similarly been around for several decades.

(192.) See 5 WIGMORE, supra note 25, [section] 1368, at 36 ("What is the theory of [cross-examination's] efficiency? ... Upon this we commonly reflect but little."). Wigmore then goes on in section 1368 to reflect just a little more. Id. [section] 1368, at 36-38. He emphasizes that cross-examination "extract[s] ... the remaining qualifying circumstances, if any, known to the witness, but hitherto [on direct] undisclosed by him," id. at 37, as well as "facts which diminish the personal trustworthiness or credit of the witness [that also] have remained undisclosed on the direct examination," id., all this with the drama of immediately succeeding direct examination and forcing the witness to supply his own refutation, id. at 38. Wigmore does not explain, however, how these extractions occur or why a witness's cross-examination should be believed any more than her direct. Wigmore also provides a series of examples of cross-examination in litigation, as well as extensive excerpts from writings of earlier evidence scholars.

(193.) Park, supra note 42, at 162. Perhaps the closest that experimental research has come is Peter Miene, Roger C. Park & Eugene Borgida, Juror Decision Making and the Evaluation of Hearsay Evidence, 76 MINN. L. REV. 683 (1992) (reporting on one of the few hearsay experiments with real eyewitnesses to a staged incident and actual cross-examination by participating attorneys).

(194.) Like the efficacy of semantic content as a deception cue generally, see supra note 106, the proposition that unanticipated questions are likely to reveal more semantic cues than anticipated questions is oddly neglected in empirical and experimental research. VRIJ, supra note 82, at 111 ("[I]t might be that spontaneous lies contain more verbal indicators of deception than planned lies. I am not aware of research investigating this issue to date.").

(195.) Clark v. Pa. R.R. Co., 328 F.2d 591,594 (2d Cir. 1964); see also United States v. Agurs, 427 U.S. 97, 108 (1976) (describing express rejection of "sporting theory of justice" as rationale for Brady rule that prosecutor must upon request turn over exonerating evidence to accused, as provided in Brady v. Maryland, 373 U.S. 83, 87 (1963)); United States v. Filani, 74 F.3d 378, 384-85 (2d Cir. 1996) ("Our court has never embraced the so-called sporting theory of the common law."); cf 6 WIGMORE, supra note 25, [section] 1845, at 488-90 (describing "sportsmanlike instincts" and the "game of litigation" in describing the old common law's aversion to discovery).

(196.) Tentative Recommendations and a Study Relating to the Uniform Rules of Evidence--Article VL Extrinsic Policies Affecting Admissibility, 6 CAL. L. REVISION COMM'N REPS. RECOMMENDATIONS & STUD. 601, 612 (1964) ("[S]urprise frequently is the essential tool for recognizing the truth.").

(197.) FED. R. EVID. 403. Other rules of evidence do require pretrial disclosure. See, e.g., FED. R. EVID. 404(b), 413, 609 (requiring disclosure of intent to use prior act evidence for various purposes, including proving state of mind, proving propensity to commit sex offenses, and impeaching a witness).

Continuances may mitigate surprise, but only partially. In the first place, continuances are unlikely to be granted when surprise is part of a strategy to catch the witness in a contradiction. See, for example, the discussion immediately below of the common law and Federal Rules procedures for impeaching a witness by her prior inconsistent statement, neither of which involve the granting of continuances. Second, continuances are not as a matter of course granted during depositions. Third, even when granted, a continuance allows the surprised party a relatively small amount of time to counterprepare.

(198.) The exception is KAY. STAT. ANN. [section] 60-445 (2004).


(200.) FED. R. EVID. 403 advisory committee's note.

(201.) See, e.g., Green Constr. Co. v. Kan. Power & Light Co., 759 F. Supp. 740, 745 (D. Kan. 1991) ("KPL next contends that it was unfairly prejudiced by the surprise introduction of two exhibits ... [that were] not produced to KPL until the Friday of the week preceding the commencement of trial.... The court finds that KPL's contentions in this regard are without merit. [Among other reasons,] surprise is not generally a recognized ground for excluding evidence under the Federal Rules of Evidence.").

(202.) FED. R. EVID. 613 advisory committee's note to subdivision (a).

(203.) WEINSTEIN & BERGER, supra note 104, [section] 613.0212][b] ("Rule 613 gives greater weight to surprise than to warning as a technique for ferreting out the truth.").

(204.) The Queen's Case, 129 Eng. Rep. 976 (K.B. 1820).

(205.) See, e.g., WEINSTEIN & BERGER, supra note 104, [section] 403.0214][b] ("The development of discovery practice and pretrial procedure has reduced the possibility of surprise in many civil cases."); see also Park, supra note 42, at 165-66 (discussing the impact of depositions on cross-examination).

(206.) The extent of discovery is sensitive to the size of the stakes of the case. Most cases are apparently too small to warrant extensive discovery. James S. Kakalik, Deborah R. Hensler, Daniel McCaffrey, Marian Oshiro, Nicholas M. Pace & Mary E. Vaiana, Discovery Management. Further Analysis of the Civil Justice Reform Act Evaluation Data, 39 B.C. L. REV. 613,637-38, 643 (1998); Thomas E. Willging, Donna Stienstra, John Shapard & Dean Miletich, An Empirical Study of Discovery and Disclosure Practice Under the 1993 Federal Rule Amendments, 39 B.C. L. REV. 525, 527, 532 (1998); see also JAMES S. KAKALIK, TERENCE DUNWORTH, LAURAL A. HILL, DANIEL MCCAFFREY, MARIAN OSHIRO, NICHOLAS M. PACE & MARY E. VAIANA. AN EVALUATION OF JUDICIAL CASE MANAGEMENT UNDER THE CIVIL JUSTICE REFORM ACT (1996).

(207.) See, e.g., FED. R. CIV. P. 26(b)(3) (stating that documents and tangible things prepared in anticipation of litigation are discoverable only on showing of substantial need and practical inability to obtain materials by other means; even when discovery of these materials is allowed, mental impressions, conclusions, opinions, and legal theories are still protected); Hickman v. Taylor. 329 U.S. 495, 511-13 (1947) (explicitly protecting against disclosure of mental impressions, etc., even if not in connection with discovery of documents and tangible things; codified in part in Federal Rule of Civil Procedure 26(b)(3)).

(208.) Contrary to this conditional, the leading treatise on evidentiary tactics suggests that discovery can itself be a tool in creating trial surprise. Surprise, after all, is a matter of upsetting expectations. And formal discovery is the perfect setting in which to create expectations that can later be dashed at trial. EDWARD J. IMWINKELRIED & DAVID A. SCHLUETER, FEDERAL EVIDENCE TACTICS [section] 5.01 (2004).

(209.) This would be the ease, for example, if they were inadmissible hearsay. See, e.g., FED. R. EVID. 801-807 (defining hearsay and exceptions). This, of course, assumes that the distinction between substantive and impeachment use is clear and operable to the factfinder. Limiting instructions may be somewhat helpful in this regard. See, e.g., FED. R. EVID. 105. Furthermore, the distinction may have real effect in determining whether there is sufficient substantive evidence to send the ease to the jury.

(210.) Cf. Park, supra note 42, at 165.

(211.) See, e.g., FED. R. EVID. 801(d)(1)("Prior statement by witness").

(212.) See, e.g., FED. R. EVID. 801(d)(2)("Admission by party-opponent").

(213.) FED. R. CIV. P. 56(e) (stating that deposition transcripts may supplement supporting or opposing affidavits). But note that in ruling on summary judgment, the court will consider only such deposition testimony as would be admissible at trial were it there recited. Visser v. Packer Eng'g Assocs., Inc., 924 F.2d 655,659 (7th Cir. 1991) (holding that proffered affidavits are not to be considered in ruling on summary judgment because the statements therein do not comport with the personal knowledge requirements for lay witnesses under Federal Rule of Evidence 602).

(214.) See, e.g., Symposium, The Vanishing Trial, 1 J. EMPIRICAL LEG. STUD. 459 (2004).

(215.) One might counter-counterrespond here that required initial disclosures, see FED. R. CIV. P. 26(a), and requests to produce documents and things, see FED. R. CIV. P. 34, 45, may give the deponent (via her proponent) important clues about the content of her deposition. But, in fact, this is not so clear. First, such devices also provide the other side with fodder for surprising the witness. Thus, the deponent William H. Gates was, in a sense, surprised by e-mails that Microsoft's own lawyers had turned over to the Justice Department. Second, with respect to required initial disclosures, the federal rules treat information used solely for impeachment differently. Federal Rule of Civil Procedure 26(a)(1)(A) exempts from required "initial disclosures" the contact information of individuals likely to have discoverable information if that information is to be used solely for impeachment. Federal Rule of Civil Procedure 26(a)(1)(B) treats initial disclosures of documents and tangible things similarly. And Federal Rule of Civil Procedure 26(a)(3) extends the same treatment to "pretrial disclosures" required to be made a month before trial. (That said, much information useful for impeachment might have a substantive use as well, and is thereby not protected from required disclosure by these provisions.) Third, the judge often has great discretion under rules like Federal Rule of Civil Procedure 16 regarding how to schedule discovery. And she will often take her cues from the parties. This affords some leeway in structuring discovery so as to maximize deponent surprise when that is helpful in separating truthful from untruthful testimony.

(216.) Miranda v. Arizona. 384 U.S. 436 (1966) (requiring procedural safeguards during in-custody interrogation to render effective suspect's privilege against self-incrimination, establishing right to counsel prior to questioning and during interrogation, and allowing for waiver of these rights), qualified by, e.g., Illinois v. Perkins, 496 U.S. 292 (1990) (finding Miranda not applicable where suspect makes voluntary statement to someone that he does not realize is a law enforcement officer), and New York v. Quarles, 467 U.S. 649 (1984) (discussing "public safety" exception).

(217.) See. e.g., FED. R. EVID. 801(d)(2) ("Admission by party-opponent").

(218.) See, e.g., FED. R. EVID. 613 (implicitly allowing impeachment by prior inconsistent statement); FED. R. EVID. 801(c) (limiting hearsay prohibition to statements "offered ... to prove the truth of the matter asserted"); Harris v. New York, 401 U.S. 222 (1971) (holding exclusionary rule not applicable to impeachment use of custodial statement, even if statement goes to historical merits of the case and is not "collateral"). With regard to the impeachment use of silence, see Jenkins v. Anderson, 447 U.S. 231, 240-41 (1980) ("[U]se of prearrest silence to impeach a defendant's credibility does not violate the Constitution.") (emphasis added). But see Doyle v. Ohio, 426 U.S. 610 (1976) (holding that post-Miranda warning silence may not be used to impeach the accused).

(219.) For example, although post-Miranda warning silence may not be used to impeach the accused, Doyle, 426 U.S. 610, this protection is inapplicable to nonparty witnesses.

(220.) Brady v. Maryland, 373 U.S. 83, 87 (1963): see also Seth F. Kreimer & David Rudovsky, Double Helix, Double Bind: Factual Innocence and Postconviction DNA Testing, 151 U. PENN. L. REV. 547, 577-87 (2002) (discussing Brady and its progeny in arguing for a right to postconviction DNA evidence).

(221.) See, e.g., FED. R. CRIM. P. 12.1, 16, 26.2 (provisions touching on discovery).

(222.) This topic is further discussed infra at Part IV.B.2.

(223.) See, e.g., FED. R. CIV. P. 31.


(225.) See, e.g.. FED. R. CIV. P. 26-37.

(226.) See, e.g., FED. R. CRIM. P. 16.

(227.) See, e.g.. FED. R. CIV. P. 34 (concerning requests from parties in civil cases); FED. R. CIV. P. 45(a)(1)(C) (concerning requests from any person in a civil case).

(228.) Sue, e.g., FED. R. CIV. P. 16(c)(6).

(229.) See. e.g., FED. R. CIV. P. 16(c)(3) (stating that obtaining admissions of fact is one objective of the pretrial conference); FED. R. CIV. P. 36 (concerning requests for admissions).

(230.) See, e.g., FED. R. CIV. P. 30.

(231.) See supra note 227.

(232.) Under the current federal rules for civil cases, this would require stipulation by the parties or special authorization by the court. FED. R. CIV. P. 30(a)(2)(B), 33(a).

(233.) Naturally, the desire to exploit cognitive limitations must be balanced against the desire to limit the expense of litigation, inclusive of the desire to induce settlement.

(234.) For more on this comparison, see infra Part IV.B.4.

(235.) The questioner may indeed repeat earlier questions: there is no "asked and answered" objection on deposition. FED. R. CIV. P. 26(b).

(236.) Holding, supra note 8, at 152 ("It appears that the visual and auditory tasks showing declines in sensitivity are those which present a high event rate, with stimuli occurring every 2 or 3 seconds, and which demand an ability to make rapid perceptual comparisons involving memory."); id. (reviewing "Cambridge Cockpit" studies and noting that "[a]ttention began to be reserved for items of central importance, like the course heading and speed indicators, while peripheral items like the fuel gauge were neglected"); id. at 159-60 (reviewing COPE tests, noting that "[w]e can be virtually certain that, other things being equal, fatigued subjects will choose to exert less effort" and that this implies "a tendency toward carelessness or 'cutting comers'").

(237.) FED. R. CIV. P. 30(d)(2), for example, was amended in 2000 to limit a deposition upon oral questions to one seven-hour day, unless the court otherwise orders or the parties otherwise stipulate. See generally CHARLES ALAN WRIGHT, ARTHUR R. MILLER & RICHARD L. MARCUS, FEDERAL PRACTICE AND PROCEDURE [section] 2104.1 (Supp. 2004) (describing deposition time limits in more detail).

(238.) Another important aspect of the performance of testimony--one with less connection to either the positive role of cognitive limits or the concept of "liveness"--is the fact that witnesses are generally forced to answer. In ordinary social interaction, those who face questions with damaging answers can often avoid the dilemma of either sustaining the damage or floating a lie by pretending to misunderstand the question or feigning a good reason to avoid it. In all this, they may perhaps count on the questioner's feeling constrained from following up by norms of social interaction. VRIJ, supra note 82, at 222-23. Such tactics are largely infeasible at trial or on deposition, where the social norm is adversarial, where it is not up to the witness whether she will answer the question, where the questioner almost always has both the motivation and the opportunity to follow up, and where, generally, the subpoena and contempt power of the court imply that a witness who simply refuses to take the test of testimony may find herself in jail or subject to a substantial fine. See, e.g., 18 U.S.C. [section] 401 (2000 & Supp. II 2002) (criminal contempt); FED. R. CIV. P. 45 (subpoena in civil cases); FED. R. CRIM. P. 17 (subpoena in criminal cases).

(239.) See sources cited supra note 10.

(240.) Conventional wisdom and game theoretic analysis tend to emphasize a "first-mover advantage": one that derives from the ability to commit to an action before one's opponent makes her choice. For a lucid discussion of this issue, see Kyle Bagwell & Asher Wolinsky, Game Theory and Industrial Organization, in 3 HANDBOOK OF GAME THEORY WITH ECONOMIC APPLICATIONS [section] 2 (Robert J. Aumann & Sergiu Halt eds., 2002).

(241.) To be precise, I refer here to rational actors who are cognizant of the full state space. If the state space is not fully know, n, open-loop and closed-loop decisionmaking are not equivalent, even for a rational actor: the actor could not, in fact, plan ex ante for all contingencies when he cannot conceive of all contingencies. For an analysis of rational decisionmaking when the state space is not fully known, see Itzhak Gilboa & David Schmeidler, Case-Based Decision Theory, 110 Q.J. ECON. 605 (1995).

(242.) Cf. JOHN VON NEUMANN & OSKAR MORGENSTERN, THEORY OF GAMES AND ECONOMIC BEHAVIOR 79-84 (3rd ed. 1953) (defining "strategy" in extensive-form games).

(243.) The comparison here drawn should be contrasted with the separate issue of whether using work product to refresh a witness's memory waives the protection. See, e.g., Alfreda Robinson, Duet or Duel: Federal Rule of Evidence 612 and the Work Product Doctrine Codified in Civil Procedure Rule 26(b)(3), 69 U. CIN. L. REV. 197 (2000).

(244.) Federal Rule of Evidence 612 applies only to witnesses. This discussion assumes there is no waiver and that the notes are not used by the opposing side's own witnesses.

(245.) FED. R. CIV. P. 26(b)(3); Hickman v. Taylor, 329 U.S. 495, 511-13 (1947).

(246.) FED. R. CIV. P. 26(b)(3) (stating that even when ordering discovery of materials prepared in anticipation of litigation, "[the] court shall protect against disclosure of the mental impressions, conclusions, opinions, or legal theories of an attorney"); Hickman, 329 US. at 511-13.

(247.) Michael E. Tigar, Habeas Corpus and the Penalty of Death, 90 COLUM. L. REV. 255, 256 (1990) (book review).

(248.) See Summation Legal Techs., Inc., Summation Legal Technologies, Inc.: America's #1 Litigation Support Software, at (last visited Nov. 10, 2004).

(249.) See inData Corp., Welcome to inData Corporation, at http:// (last visited Nov. 10, 2004).

(250.) Summation Legal Techs., Inc., Globally Searching Your Case: Summation Core Simplicity, at (last visited Nov. 10, 2004).

(251.) Id. (describing "Case Organizer").

(252.) The aforementioned Julian case would seem to imply that work product protection would not bar disclosure. Much of the content of the binders in Julian was available to the other side. It was the particular choice and organization of these materials in the binders that endowed the binders with work product protection. Nevertheless, the court ordered disclosure under Federal Rule of Evidence 612. James Julian, Inc. v. Raytheon Co., 93 F.R.D. 138, 144 (D. Del. 1982). It would be difficult to argue that a software program that organizes and presents electronic data should be treated differently.

To be sure, there may be some question as to whether computer data is a "writing" for purposes of Federal Rule of Evidence 612. But even if it is deemed not to be, similar disclosure procedures would probably apply. Note in this regard that Rule 612 applies only to "writings" also with respect to materials used while testifying. And yet it is well established that procedures similar to those laid out in Rule 612 apply to any material used to refresh recollection on the stand. See, e.g., Baker v. State, 371 A.2d 699, 703 (Md. Ct. Spec. App. 1977) ("The opposing party, of course, has the right to inspect the memory aid, be it a writing or otherwise, and even to show it to the jury.") (emphasis added).

(253.) Steve Hamm, A Database to Die For, BUS. WK., Mar. 15, 1999, at 40 (emphasis added).

(254.) THOMAS A. MAUET, TRIAL TECHNIQUES 13-29 (5th ed. 2000).

(255.) Id. at 23.

(256.) Id. at 24.

(257.) Id.; see also id. at 25 (suggesting the use of "memorable words and phrases that encapsulate the essence of your case" and "tags you put on people, events, and things ... [to] convey images"); id. at 26 (advising readers to emphasize the "human element," e.g., motivations); id. at 26 (advising the use of "gripping visual aids"); id. at 26-27 (advising tapping into the instinct of storytelling to "organize, humanize, and dramatize"); id. at 28 (stating that "an advocate grabs and holds the jurors' attention").

(258.) Id. at 406-07.

(259.) Id. at 409-10.

(260.) Id. at 407.

(261.) See generally Christopher N. May, "What Do We Do Now?": Helping Juries Apply the Instructions, 28 LOY. L.A. L. REV. 869 (1995) (reviewing commentary and law on the judge's ability to summarize evidence for the jury).

(262.) Quercia v. United States, 289 U.S. 466, 469 (1933) ("In charging the jury, the trial judge is not limited to instructions of an abstract sort. It is within his province, whenever he thinks it necessary, to assist the jury in arriving at a just conclusion by explaining and commenting upon the evidence, by drawing their attention to the parts of it which he thinks important; and he may express his opinion upon the facts, provided he makes it clear to the jury that all matters of fact are submitted to their determination."); United States v. Filani, 74 F.3d 378, 385 (2d Cir. 1996) ("[The] trial court may actively participate and give its own impressions of the evidence or question witnesses, as an aid to the jury, so long as it does not step across the line and become an advocate for one side.").

(263.) WEINSTEIN & BERGER, supra note 104, [section] 107.03[2].

(264.) The chief case offered in support of this proposition in Weinstein's Federal Evidence, Geders v. United States, 425 U.S. 80 (1976), seems of questionable relevance. Weinstein's Federal Evidence also cites Quercia, 289 U.S. 466. But Quercia talks about these powers of the judge "[i]n charging the jury." And when the court in that case says that it is "within [the judge's] province, whenever he thinks it necessary, to assist the jury in arriving at a just conclusion by explaining and commenting upon the evidence," it is unclear whether "whenever" means "whenever during the trial of a particular case" or "in whichever case, at the end thereof." See Quercia, 289 U.S. at 469 (emphasis added).

(265.) WEINSTEIN & BERGER, supra note 104, [section] 107.13[1].

(266.) United States v. Maclean, 578 F.2d 64, 64-65 (3d Cir. 1978) ("[The] unanimous view of federal appellate courts that have decided the issue ... [w]hether or not to allow note-taking by jurors is [that it is] a matter committed to the sound discretion of trial judges.... At least one state has a rule prohibiting note-taking. Most states, however, leave the question to the discretion of the trial judge.").

(267.) In the Eastern District of Pennsylvania, for example, of the thirteen judges who indicate either way in their "Judge's Procedures" whether they permit jury note-taking, nine say they do, three decide on a case-by-case basis, and one says she generally does not. (This tally does not include senior judges.) See links compiled at U.S. District Court, Eastern District of Pennsylvania, Clerk's Office, Judges' Procedures, at (last visited Nov. 11,2004).

(268.) Judge Stewart Dalzell--Policies and Procedures (Part 1), at (last visited Nov. 11, 2004) ("[J]urors have availed themselves of that opportunity [to take notes] in every case.").

(269.) Maclean, 578 F.2d at 66.

(270.) See e.g., FED. R. EVID. 803(6).

(271.) See e.g., FED. R. EVID. 803(8).

(272.) See e.g., FED. R. EVID. 803(9).

(273.) See e.g., FED. R. EVID. 803(17).

(274.) See e.g., FED. R. EVID. 803(18).

(275.) See e.g., FED. R. EVID. 801(d)(2).

(276.) See e.g., FED. R. EVID. 801(d)(1).

(277.) See e.g., Chief Judge James T. Giles, Policies and Procedures, at (last visited Nov. 18, 2004) ("[In civil cases, i]f the jurors are specific in their request to have testimony read back, Judge Giles will arrange to have it read. He will allow the replaying of tape recordings and videotapes if requested by the jury.").

(278.) See, e.g., United States v. Silverstein, 732 F.2d 1338, 1347 (7th Cir. 1984) ("The indictment ... [had been] given to the jury with the usual instruction that it was not evidence....").

(279.) Before being apprehended, Gugasian was known as the "Friday Night Bank Robber" for his proclivity to strike at closing time on Friday. Joseph A. Slobodzian, The Mask Comes Off, PHILA. INQUIRER, Jan. 4, 2004, at M1.

(280.) Id. ("[Gugasian is] without doubt the most prolific, successful bank robber in U.S. history: [he committed] scores of heists, all on Fridays, going back three decades, netting him about $2 million."); see Bob Laylo, Thief Who Rivals Dillinger for Most Bank Heists Is Jailed, ALLENTOWN MORNING CALL (Allentown, Pa.), Dec. 10, 2003, at A1 ("'Gugasian may well be the most prolific bank robber this nation has ever known,' said Assistant U.S. Attorney Linwood C. Wright Jr."). Of course, there is always the possibility that another bank robber, yet to be apprehended, is more successful. Furthermore, the bare number of unsolved robberies says nothing about the average take.

(281.) Slobodzian, supra note 279.

(282.) Id.

(283.) Laylo, supra note 280 (quoting retired Pennsylvania state trooper John Mauro, who stated that Gugasian was "one of the most clever criminals l investigated" and that he was "one of your true professionals").

(284.) Id. (attributing this assessment to FBI agent Raymond J. Carr).

(285.) Id. (map reading); Slobodzian, supra note 279 (weapons, survival, self-defense).

(286.) Laylo, supra note 280 (stating that Gugasian frequently jogged in street clothes wearing a backpack); Slobodzian, supra note 279 ("Health-food fanatic, devotee of yoga and meditation, third-degree black belt in karate, Gugasian is a lean, muscular 5-foot-9 in superb condition.").

(287.) Slobodzian, supra note 279.

(288.) Id.

(289.) Id.

(290.) Laylo, supra note 280.

(291.) Id.; Slobodzian, supra note 279.

(292.) Martha Stewart and Peter Bacanovic were convicted on March 5, 2004. Sentencing was scheduled for June 17, 2004. Hays & Eaton, supra note 28; see also Superseding Indictment, United States v. Stewart, 323 F. Supp. 2d 606 (S.D.N.Y. 2004) (S1 03 Cr. 717 (MGC)), available at (charging, inter alia, violations of 18 U.S.C. [section] 1621 (2000) (false declarations before grand jury or court, including ancillary proceedings) and 18 U.S.C. [section] 1505 (2000) (obstruction of proceedings before departments, agencies, and committees)). Although Stewart has started serving her sentence, her appeal is still pending. Constance Hays, Martha Stewart Wants to Start Sentence Early, N.Y. TIMES, Sept. 16, 2004, at A1; Barry Meier, Martha Stewart Assigned to Prison in West Virginia, N.Y. TIMES, Sept. 30, 2004, at C1. Notwithstanding Stewart's appeal, I will assume for purposes of argument that the government's account of events is accurate.

There is also a separate civil action against Stewart and Bacanovic. Complaint, SEC v. Stewart, No. 03 CV 4070 (NRB) (S.D.N.Y. filed June 4, 2003) (suing for civil fines and injunctive relief under [section] 17(a) of the Securities Act of 1933, 15 U.S.C. [section] 77q(a) (2000); [section] 10(b) of the Securities Exchange Act of 1934, 15 U.S.C. [section] 78j(b) (2000); and Rule 10b-5, 17 C.F.R. [section] 240.10b-5 (2002), thereunder), available at

(293.) His daughter also did the same. Complaint, SEC v. Stewart, at 1.

(294.) Id. at 1, 2.

(295.) Superseding Indictment, United States v. Stewart, at 7 ("On December 27, 2001, at approximately 10:04 a.m. (EST), within minutes after being informed of the sale and attempted sale of the Waksal Shares, PETER BACANOVIC called MARTHA STEWART. After being told that STEWART was in transit and unavailable, BACANOVIC left a message, memorialized by STEWART's assistant, that 'Peter Bacanovic thinks ImClone is going to start trading downward.'").

(296.) In addition, the ImClone CEO's phone log for later that same day reads, "Martha Stewart something is going on with ImClone and she wants to know what...." Samuel Waksal's December 27, 2001 Message Log,

(297.) Superseding Indictment, United States v. Stewart, at 7.

(298.) Id. at 16.

(299.) Id. ("[One week before the phone message, Bacanovic] printed a 'worksheet' that listed each of the stocks held by Martha Stewart at Merrill Lynch, including ImClone.... Bacanovic made handwritten notes in blue ballpoint ink on the Worksheet concerning transactions and planned transactions in Stewart's account.... Bacanovic made no notes on the Worksheet regarding any purported decision to sell Stewart's ImClone shares at $60 per share."); see also infra note 304 (discussing the allegation that Bacanovic later penned in "@60").

(300.) This is again related to the problem of coordination. See supra notes 31, 175, and 183.

(301.) Slobodzian, supra note 279 (noting this point, and also showing a photograph of readily intelligible notes listing bank names and their hours).

(302.) Under the "best evidence rule," as codified, for example, in the Federal Rules of Evidence, "[t]o prove the content of a writing, recording, or photograph, the original ... is required...." FED. R. EVID. 1002. However, the general rule contains an exception for, inter alia, a "duplicate," which is "admissible to the same extent as an original unless a genuine question is raised as to the authenticity of the original or in the circumstances it would be unfair to admit the duplicate in lieu of the original." FED. R. EVID. 1003. "Duplicate" is defined in Federal Rule of Evidence 1001 to include the products of the forms of reproduction mentioned in the text.

(303.) Superseding Indictment, United States v. Stewart, at 11-12 ("On or about January 31, 2002, after learning that the FBI and the U.S. Attorney's Office had requested an interview with her, and immediately following a lengthy conversation with her attorney, MARTHA STEWART accessed the phone message log maintained on computer by her assistant and reviewed the phone message that PETER BACANOVIC had left for her on December 27, 2001. In furtherance of the conspiracy, and knowing that BACANOVIC's message for STEWART was based on information regarding the sale and attempted sale of the Waksal Shares that BACANOVIC subsequently caused to be conveyed to her, STEWART deleted the substance of BACANOVIC's phone message, changing the message from 'Peter Bacanovic thinks ImClone is going to start trading downward,' to 'Peter Bacanovic re imclone.'").

(304.) Id. at 17 ("BACANOVIC added the notation '@ 60' near the entry for ImClone."). To be sure, Bacanovic was acquitted of the charge of making and using false documents. Moreover, the government's expert on ink analysis has been indicted for perjury in connection with his testimony regarding the worksheet. However, the defense's ink expert agreed that "@ 60" was in a different ink from the rest of the worksheet. (There remains some issue, however, as to whether the rest of the worksheet was in the same ink.) Letter from David N. Kelly, United States Attorney, Southern District of New York, to the Honorable Miriam Goldman Cederbaum, Federal District Judge, Southern District of New York 2 (May 21, 2004), available at

(305.) Superseding Indictment, United States v. Stewart, at 12 ("After altering the message, STEWART directed her assistant to return the message to its original wording.").

(306.) Id. at 17 ("BACANOVIC altered the Worksheet, using ink that was blue ballpoint, but was scientifically distinguishable from the ink used elsewhere on the Worksheet.").

(307.) Many corporations have "document retention policies" to similar effect. See Sanchirico, supra note 31, at 1273.

(308.) Id. at 1271.

(309.) Indeed, a similar story applies to evidentiary emissions in the form of eyewitness memory.

Chris William Sanchirico, Professor of Law, Business, and Public Policy, University of Pennsylvania Law School and Wharton School; Visiting Professor of Law, University of Chicago Law School, Autumn 2004. This Article has benefited from the comments and questions of workshop participants at the American Law and Economics Association's 2004 Annual Meeting, Fordham University School of Law, the New York University Economics Department, Northwestern University School of Law, Stanford Law School, the University of California, Berkeley School of Law (Boalt Hall), and the University of Chicago Law School, as well as those of Jonathan Bendor, Vincent Blasi, Craig Callen, Aaron Edlin, Alev Erisir, George Fisher, Todd Henderson, Seth Kreimer, Thomas Miles, Roger Park, Mario Rizzo, Susan Rose-Ackerman, Peter Siegelman, David Skeel, David Sklansky, Eleanor Swift, and John Sylla. Special thanks to Charles Goetz and Daniel Ortiz. Katherine Bierlein and Penn law librarians William Mulherin, Clayton Garthwaite, and Merle Slyhoff provided invaluable research assistance.
COPYRIGHT 2004 Stanford Law School
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Sanchirico, Chris William
Publication:Stanford Law Review
Date:Nov 1, 2004
Previous Article:The People Themselves: Popular Constitutionalism and Judicial Review.
Next Article:Reform(aliz)ing copyright.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters