Juridical proof and the best explanation

Law and Philosophy

Volumn 27, Issue 3, 2008, Pages 223-268

  Pardo, Michael S ^a,f   Allen, Ronald J ^b,c,d,e  

^a UNIVERSITY OF ALABAMA   (United States)

^b Northwestern School of Law

^c Procedural Law Research Center

^d CHINA UNIVERSITY OF POLITICAL SCIENCE AND LAW   (China)

^e Searle Center

^f UNIVERSITY OF ALABAMA SCHOOL OF LAW   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 41449101021     PISSN: 01675249     EISSN: None     Source Type: Journal    
DOI: 10.1007/s10982-007-9016-4     Document Type: Article

References (191)

1. The classic article coining the phrase and explaining the idea is Gilbert Harman, The Inference to the Best Explanation,' Philosophical Review 74 (1965): 88-95;
2. see also Peter Lipton, Inference to the Best Explanation (London: Routledge, 2004);
3. William G. Lycan, Judgment and Justification (Cambridge University Press, 1988);
4. Paul R. Thagard, 'Evaluating Explanations in Law, Science, and Everyday Life,' Current Directions in Psychological Science 15 (2006): 141-45;
5. Brian Leiter, 'Moral Facts and Best Explanations,' Social Philosophy & Policy (2001): 79-101;
6. Timothy Day and Harold Kincaid, 'Putting Inference to the Best Explanation in its Place,' Synthese 98 (1994): 271-95;
7. Yemima Ben-Menahem, 'The Inference to the Best Explanation,' Erkenntnis 33 (1990): 209-24;
8. Paul R. Thagard, 'The Best Explanation: Criteria for Theory Choice,' Journal of Philosophy 75 (1978): 76-92.
9. For recent examples see Dale A. Nance and Scott B. Morris, 'Juror Understanding of DNA Evidence: An Empirical Assessment of Presentation Formats for Trace Evidence with a Relatively Small Random-Match Probability,' Journal of Legal. Studies 34 (2005): 395;
10. Michael O. Finkelstein and Bruce Levin, 'On the Probative Value of Evidence from a Screening Search,' Jurimetrics Journal 43 (2003): 265-266;
11. Deborah Davis and William C. Follette, 'Toward an Empirical Approach to Evidentiary Ruling,' 27 Law and Human Behavior 27 (2003): 661;
12. Richard D. Friedman and Roger C. Park, 'Sometimes What Everybody Thinks They Know is True,' 27 Law and Human Behavior 27 (2003): 629;
13. Deborah Davis and William C. Follette, 'Rethinking the Probative Value of Evidence: Base Rates, Intuitive Profiling, and the "Postdiction" of Behavior,' Law and Human Behavior 27 (2002): 133, 156;
14. Dale A. Nance and Scott B. Morris, 'An Empirical Assessment of Presentation Formats for Trace Evidence with a Relatively Large and Quantifiable Random Match Probability,' Jurimetrics Journal 42 (2002): 403.
15. For earlier examples see John Kaplan, 'Decision Theory and the Factfinding Process,' Stanford Law Review 20 (1968): 1065;
16. Michael Finkelstein and William Fairley, 'A Bayesian Approach to Identification Evidence,' Harvard Law Review 83 (1970): 489;
17. Richard Lempert, 'Modeling Relevance,' Michigan Law Review 75 (1977): 1021;
18. Richard D. Friedman, 'Route Analysis of Credibility and Hearsay,' Yale Law Journal 96 (1987): 667;
19. Peter Tillers and David A. Schum, 'Hearsay Logic,' Minnesota Law Review 76 (1992): 813.
20. The neglect we are referring to do concerns analytical attempts to illuminate the nature of various aspects of the proof process. Empirical studies of jury behavior, by contrast, have focused on explanatory-based reasoning, and they have concluded that jurors do indeed rely primarily on explanatory criteria in deciding cases. See Nancy Pennington and Reid Hastie, 'A Cognitive Model of Juror Decision Making: The Story Model,' Cardozo Law Review 13 (1991): 519. Thus our discussion has the salutary effect of bring these two areas closer together.
21. For discussions of abductive reasoning in the law see David A. Schum, 'Species of Abductive Reasoning in Fact Investigation in Law,' Cardozo Law Review 22 (2001): 1645;
22. see Kola Abimbola, 'Abductive Reasoning in Law: Taxonomy and Inference to the Best Explanation,' Cardozo Law Review 22 (2001): 1683;
23. John R. Josephson, 'On the Proof Dynamics of Inference to the Best Explanation,' Cardozo Law Re view 22 (2001) 1621.
24. Limitations on probability-based approaches to explain the probative value of evidence are discussed in Ronald J. Allen and Michael S. Pardo, 'The Problematic Value of Mathematical Models of Evidence,' Journal of Legal Studies 36 (2007): 107.
25. See Fed. R. Evid. 401, 403.
26. See Fed. R. Civ. P. 56.
27. See Fed. R. Civ. P. 50.
28. See Fed. R. Civ. P. 50(b).
29. See Fed. R. Crim P. 29; see also Jackson v. Virginia, 443 U.S. 307, 324 (1979) (articulating standard as whether "[n]o rational trier of fact could have found proof of guilt beyond a reasonable doubt.")
30. If all men are mortal, and if Socrates is a man, then Socrates is mortal.
31. But various aspects of the process may involve deductions once the jury finds certain propositions to be true. For example, if a jury is told that the defendant was negligent if he drove through a red light, and the jury finds that the defendant did indeed drive through a red light, then they will (or ought) to deduce that the defendant was negligent.
32. The fact that such conclusions are not guaranteed, but may be seen as more or less likely, is what invites probability-based approaches to the legal process, as it does in other areas. See generally Probability is the Very Guide of Life: The Philosophical Uses of Chance (Henry E. Kyburg, Jr. and Marian Thalos eds.) (Peru, Ill.: Open Court Press, 2003).
33. Peter Lipton, supra note 1 at 9. Enumerative induction may also involve inferences to general conclusions, e.g., that all or most As are B.
34. But it need not be-one may infer, for example, mathematical, definitional, conceptual, or grammatical explanations of given premises.
35. Inductive inferences may often be recast either as abductive or enumerative. Gilbert Harman has suggested that all inductive inferences involve abduction. Harman, supra note 1 at 88-95. Richard Fumerton, by contrast, has argued that all abduction may be described as an enumerative inference. R. A. Fumerton, 'Induction and Reasoning to the Best Explanation,' Philosophy of Science 47 (1980): 589-600. Consider a fact finder's inference from the fact that the defendant confessed to the conclusion that the defendant is guilty. This inference could be explicated as abductive (that the defendant is guilty best explains why he confessed) or as enumerative (most people in the past who have confessed have been guilty; therefore, the defendant is likely guilty as well).
36. Peter Lipton, 'What Good is an Explanation?,' in Explanation: Theoretical Approaches and Explanations 56 (G. Hon and S.S. Rakover eds.) (Norwell, MA and Dordrecht: Kluwer Academic Publishers, 2001);
37. see also Peter Lipton, 'Is Explanation a Guide to Inference?,' in Explanation: Theoretical Approaches and Explanations 93 (G. Hon and S.S. Rakover eds.) (Norwell, MA and Dordrecht: Kluwer Academic Publishers, 2001) ("we sometimes decide how likely a hypothesis is to be correct in part by considering how good an explanation it would provide, if it were correct.").
38. More generally, we do not rely on any precise definition of "explanation." It is a concept of which we assume most readers have at least a basic intuitive grasp. Explanations function by answering questions such as why, what, when, where, how, and so on. For more sophisticated accounts see Peter Achinstein, The Nature of Explanation 74-102 (Oxford University Press, 1983);
39. Bas C. van Fraassen, The Scientific Image 134-57 (Oxford University Press, 1980).
40. For a discussion of the ways abductive reasoning aids in generating hypotheses see David A. Schum, 'Species of Abductive Reasoning in Fact Investigation in Law,' Cardozo Law Review 22 (2001): 1645.
41. One reason a simpler explanation may be more likely is that more complex explanations involve more auxiliary premises and background assumptions and, therefore, more places to go wrong.
42. Harman, supra note 1; Paul R. Thagard, 'The Best Explanation: Criteria for Theory Choice,' Journal of Philosophy 75 (1978): 76-92.
43. There are spirited debates among philosophers of science over how precisely to define inference to the best explanation and precisely what its criteria are. Perhaps indicative of the differences between the treatment of IBE by the philosophers and its implications for the legal system, we are largely indifferent to these debates. The philosophers are engaged in conceptual analysis, and thus appropriately (we guess) argue over the "correct" definition of IBE; our concerns are intensely practical, and thus we are interested in what variables seem to matter to human decision makers and how well they work in juridical fact finding.
44. See Edward O. Wilson, Consilience: The Unity of Knowledge (New York: Random House, 1998); Thagard, supra note 20.
45. This example is based on one involving Adam, Eve, and an apple in Morton White, A Philosophy of Culture: The Scope of Holistic Pragmatism 89-90 (Princeton University Press, 2002),
46. which in turn is based on one in H.L.A. Hart and A.M. Honore, Causation in the Law 33-34 (Oxford University Press, 1959).
47. Lipton refers to alternative possibilities as "foils," supra note 1 at 33: "A contrastive phenomenon consists of a fact and a foil, and the same fact may have several different foils. We may not explain why the leaves turn yellow in November simpliciter, but only for example why they turn yellow in November rather than in January, or why they turn yellow in November rather than blue." In Lipton's example, the foils would be "leaves turn yellow in January" and "leaves turn blue in November." In the example in the text, foils would include "this man suffered heartburn on other occasions (when he did not)" and "other people suffered heartburn (when they did not)." A contrastive explanation works by identify why the facts occurred and these foils did not. See also id. ("When I ask my, then, 3-year old son why he threw his food on the floor, he told me that he was full. This may explain why he threw it on the floor rather than eating it, but I wanted to know why he threw it rather than leaving it on his plate.").
48. But it might not if someone else could have easily planted it in the maid's pocket without her realizing it.
49. Identifying an appropriate contrast may also help to locate the meaning of statements. See Fred I. Dretske, 'Contrastive Statements,' Philosophical Review 81 (1972): 411-437. "The maid stole the necklace" (rather than the butler) when asked "Who stole the necklace?" means something different than "The maid stole the necklace" (rather than received it as a gift) when asked "Why does the maid have the necklace?" Likewise, identifying appropriate contrasts also helps to locate the reasons that would support a proposition.
50. For example (Dretske's), if one wanted to know why Clyde married Bertha, Clyde's reasons for marrying Bertha may not be the same as his reasons for marrying Bertha. Id. at 417.
51. Lipton, What Good is an Explanation?, supra note 16 at 44.
52. Id. at 44 ("Any account of understanding that rules them out is incorrect.").
53. Parties may sometimes be precluded from offering contradictory accounts. See McCormick v. Kopman, 161 N.E. 2d 720 (Ill. App. Ct. 1959).
54. Empirical evidence (and common sense) suggests that juries assume in most cases the parties have put forward the explanation that best helps their case. See Dale A. Nance, 'Naturalized Epistemology and the Critique of Evidence Theory,' Virginia Law Review 87 (2001): 1551, 1579 n. 91
55. (citing Robert H. Klonoff and Paul L. Colby, Winning Jury Trials: Trial Tactics and Sponsorship Theory (Lexis-Nexis, 1990)).
56. But that will not always be the case, for example, when parties for personal reasons don't want to reveal what actually happened even when it may offer the best explanation of the evidence and benefit them at trial. For a vivid example, see Steve Bogira, Courtroom 302: A year Behind the Scenes in an American Criminal Courthouse 236-59 (New York: Knopf, 2005), recounting a story of a 16-year-old female defendant who never revealed a prior sexual relationship with the victim, a cab driver, whom she was convicted of shooting.
57. See Ronald J. Allen, 'Factual Ambiguity and Theory of Evidence,' Northwestern University Law Review 88 (1994): 604;
58. Ronald J. Allen, 'The Nature of Juridical Proof,' Cardozo Law Review 13 (1991): 373.
59. We sometimes use the word "juror" or "jurors" to mean "fact finder," as is the case in the text at this footnote. It is obvious when we are using the term to refer specifically to jurors as compared to judges engaged in fact finding.
60. As explained more fully below, this aspect neutralizes a formal conjunction paradox in cases with multiple elements.
61. Which is simply an instantiation of how virtually everyone reasons about the world at large. See Allen, supra note 31 (discussion of scripts, narratives, etc.).
62. Pennington and Hastie, supra note 3.
63. A product of this process hopefully is a reduction or minimization of errors, and thus one can say that, in a sense, the decision goes for the story with the higher probability. However, references to probability in this context are completely epiphenomenal, or at least completely dependent upon IBE and serve no independent value.
64. Richard D. Friedman, 'Infinite Strands, Infinitesimally Thin: Story-telling, Bayesianism, Hearsay and other Evidence,' Cardozo Law Review 14 (1992): 79.
65. Assuming her interest is explaining why he suffered heartburn on this occasion. If he hadn't eaten any spicy food on this occasion (at any of its time slices), then he would not have suffered heartburn on this occasion.
66. Richard D. Friedman, '"E" is for Eclectic: Multiple Perspectives on Evidence,' Virginia Law Review 87 (2001): 2029, 2047, suggests that situations allowing for general explanations (like res ipsa loquitur) somehow pose a challenge to the theory that juries decide on the basis of the relative plausibility of competing explanations. But even quite general explanations are still explanations.
67. Although if a jury thinks that the best explanation of the evidence is that the defendant was in the back seat (and perhaps is covering for the driver), then the proper level of detail would shift and the proper contrast would be on where in the car the defendant was sitting.
68. As we discuss below, Professor Friedman presents his hypothetical as a criticism of the relative plausibility theory, which in turn rests upon IBE. As one can see, it has no negative implications for either.
69. Devices like summary judgment and judgments as a matter of law implement the general way burdens of proof accommodate these concerns. Both Richard Friedman and Dale Nance have criticized the theory of relative plausibility by suggesting that it cannot account for intermediate standards such as "clear and convincing" evidence. Friedman, Eclectic, supra note 39 at 2047;
70. Nance, supra note 30 at 1591-1592.
71. If both the prosecution and the defense offer implausible explanations of the evidence, the jury ought to acquit. Suggesting something quite similar to Ronald J. Allen, 'Rationality, Algorithms and Juridical Proof: A Preliminary Inquiry' International Journal of Evidence and Proof 1 (1997): 254, 273, Professor Josephson has proposed a definition of the reasonable-doubt standard that turns on whether there is an explanation that represents a "real possibility" of innocence.
72. See John R. Josephson, 'On the Proof Dynamics of Inference to the best Explanation,' Cardozo Law Review 22 (2001): 1621, 1642 ("A real possibility does not suppose the violation of any known law of nature, nor does it suppose any behavior that is completely unique or unprecedented, nor any extremely improbable chain of coincidences.").
73. Larry Laudan has argued (correctly) that this is no longer a process of inference to the best explanation and is rather an example of the kinds of decisions IBE was meant to avoid in the first place (i.e., inferring conclusions that are not the best explanations because they are less likely to be true). See Larry Laudan, 'Strange Bedfellows: Inference to the Best Explanation and the Criminal Standard of Proof,' (forthcoming in International Journal of Evidence and Proof). This is a necessary feature of the reasonable-doubt standard, however, and not a criticism of an explanatory account. IBE is, at root, based on the notion that explanatory success tracks likelihood of truth - the better the explanation, the more likely true. Because the criminal standard distributes errors unevenly (in favor of the defendant), it should not be surprising that the quality of the explanation needed for a pro-defendant verdict should therefore be lower.
74. As we discussed, we are not offering circular definitions. We are illuminating how explanatory factors guide the inferential processes at trial, and how the structure of the system is designed to control and foster those practices. If there is vagueness here, it inheres in the concept of "clear and convincing" evidence. How ever one wants to define that concept, it is met when an explanation is good enough to cause and justify the desired inference.
75. See New York Times v. Sullivan, 376 U.S. 254 (1964).
76. With "beyond a reasonable doubt," the system accommodates this vagueness by leaving it to juries to determine whether the standard has been met without requiring further elaboration. See Victor v. Nebraska, 511 U.S. 1, 5 (1994) ("the Constitution neither prohibits trial courts from defining reasonable doubt nor requires them to do so as a matter of course. Indeed, so long as the court instructs the jury on the necessity that the defendant's guilt be proved beyond a reasonable doubt, the Constitution does not require that any particular form of words be used in advising the jury of the government's burden of proof.") (citations omitted). For a recent critique of the Court's jurisprudence in this area see Larry Laudan, Truth, Error, and Criminal Law: An Essay in Legal Epistemology (Cambridge University Press, 2006).
77. For reasons why it cannot be see Allen and Pardo, supra note 4, and the discussion below.
78. For an example of the vagueness see Federal Civil Jury Instructions of the Seventh Circuit 35 (available at www.ca7.uscourts.gov) (defining the "clear and convincing" standard as "highly probable that it is true").
79. See United States v. Fatico, 458 F. Supp. 388, 410 (E.D. N.Y. 1978) (providing a survey of district judges on the probability they associated with various standards of persuasion - judges differed); see also Simon and Mahan, 'Quantifying Burdens of Proof: A View from the Bench, the Jury, and the Classroom,' Law and Society Review, 5 (1971): 319.
80. Thus, Professor Nance's lament, supra note 30 at 1593, is equally applicable to his own probabilistic account: "there is no clue how, even in principle, one can determine how probable the defendant's story must be in order to be plausible or in what other way a jury is to decide whether a story is plausible."
81. Federal Rule of Evidence 401 defines "relevant evidence" as "having any tendency to make the existence of any fact of consequence to the determination of the action more probable or less probable than it would be without the evidence." Explanatory considerations establish the "any tendency" aspect of evidence (the logical relevance); the substantive law determines which facts are of consequence. Parties must therefore construct explanations that include (or fail to include, if on the other side) these facts of consequence.
82. See Michael S. Pardo, 'The Field of Evidence and the Field of Knowledge,' Law & Philosophy 24 (2005): 321, 374-83.
83. See Allen, Factual Ambiguity, supra note 31.
84. Impeachment evidence, therefore, weakens the explanatory connection by offering contrary explanations for the testimony.
85. See Peter Tillers, 'If Wishes Were Horses: Discursive Comments on Attempts to Prevent Individuals from Being Unfairly Burdened by their Reference Classes,' Law, Probability and Risk 4 (2005): 33-49;
86. Mark Colyvan, Helen M. Regan, and Scott Ferson, 'Is it a Crime to belong to a Reference Class?' Journal of Political Philosophy 9 (2001): 168-81. We discuss this debate in more detail in Allen and Pardo, supra note 4.
87. van Fraassen, supra note 17. The IBE defense of scientific realism is also challenged in Arthur Fine, The Shaky Game: Einstein Realism and the Quantum Theory (Chicago University Press, 1996).
88. van Fraassen, supra note 17 at 20.
89. Id. at 19.
90. Id. at 19-20.
91. Nancy Cartwright, How the Laws of Physics Lie (Oxford University Press, 1983).
92. Id. at 91. Likewise, Larry Laudan has argued that IBE has a poor track record because several theories once accepted due to their being the best available explanation have been falsified (he gives several examples such as the humoral theory of medicine, the theory of electromagnetic aether, and the phlogiston theory of chemistry). See Larry Laudan, 'A Confutation of Convergent Realism,' Philosophy of Science 48 (1981): 19, 33. But, again, he is discussing the existence of scientifically posited entities: "A theory's success is no warrant for the claim that all or most of its central terms refer." Id. at 47. Whatever else is going on in litigation, it is typically the case that most of its terms refer to verifiable entities.
93. Perhaps litigation over the nature of evolution and "intelligent design" is an exception. We do not mean to suggest that IBE is problematic in science only in cases involving unobservable entities and laws, but this is the context in which it appears to have been most notoriously challenged. Rather, our point is that the failure of IBE in some scientific contexts does not necessarily mean that it is ill-suited to explain legal phenomena.
94. Lipton, supra note 1 at 209; Ilkka Niiniluoto, 'Defending Abduction,' Philosophy of Science 66 (1999): S436. Empirical evidence suggests that jurors do quite well employing this strategy at trial, see Pennington and Hastie, supra note 3, even in cases involving scientific evidence.
95. See Neil Vidmar, 'The Performance of the American Civil Jury: An Empirical Perspective,' Arizona Law Review 40 (1998): 894, 898 ("Juries in medical malpractice trials, frequent targets of critics, tend to render decisions that are consistent with independent assessments of health care providers.").
96. See Lipton, supra note 1 at 142-63.
97. See Karl Popper, Realism and the Aim of Science (1983);
98. Deborah G. Mayo, 'Evidence as Passing Severe Tests: Highly Probable vs. Highly Probed Hypotheses,' in Scientific Evidence: Philosophical Theories and Applications 95-127 (Peter Achinstein ed.) (Baltimore: The Johns Hopkins University Press, 2005).
99. Similar analysis should apply in the criminal context, with a standard of beyond a reasonable doubt, and in civil cases with a "clear and convincing evidence" standard. In criminal cases, the government's explanation must be sufficiently probed such that the plausible explanations consistent with innocence have been refuted. Likewise, in cases requiring clear and convincing evidence, a plaintiffs explanation must be sufficiently probed that it weakened the potential, plausible explanations favoring the defendant such that the plaintiffs explanation is sufficiently more plausible than those favoring the defendant.
100. See, e.g., the articles cited in supra note 2.
101. A common approach to assigning probabilities to individual items of evidence is through the notion of a "likelihood ratio," that is, the probability of evidence (e.g., a confession) given a hypothesis (e.g., the defendant is guilty) divided by the probability of that evidence given the negation of the hypothesis. See Nance and Morris, supra note 2; Finkelstein and Levin, supra note 2; David H. Kaye and Jonathan Koehler, 'The Misquantification of Probative Value,' Law and Human Behavior 27 (2003): 645. For critiques of the likelihood-ratio approach see Pardo, supra, supra note 54; Allen and Pardo, supra note 4.
102. Interpreting standards of persuasion in terms of cardinal probabilities also accords with how some judges think of them. See United States v. Shonubi, 895 F. Supp. 460, 471 (E.D. N.Y. 1995), reversed on other grounds, 103 F.3d 1085 (2d Cir. 1997) ("Judges-who deal with burdens of proof on a daily basis-are inclined to think of those burdens in probabilistic terms. A survey of judges in the Eastern District of New York found general agreement that 'a preponderance of the evidence' translates into 50 + percent probability. Eight judges estimated 'clear and convincing' as between 60 and 70 percent probable (while two found this standard unquantifiable). Estimates for 'beyond a reasonable doubt' ranged from 76 to 90 percent, with 85 percent the modal response.").
103. See, e.g., Nance and Morris, supra note 2; Davis and Follette, supra note 2.
104. The approaches also diverge on how they define probative value, either as the likelihood ratio or as the difference between prior and posterior probabilities. Compare Nance and Morris, supra note 2 with Davis and Follette, supra note 2.
105. Nance and Morris, supra note 2 (DNA random-match evidence); Finkelstein and Levin, supra note 2 (match of carpet fibers); Davis and Follette, supra note 2 (infidelity as evidence of spousal murder). We critique each of these attempts in Allen and Pardo, supra note 4.
106. See, e.g., Richard D. Friedman, 'Character Impeachment Evidence: Psycho-Bayesian [!?] Analysis and a Proposed Overhaul,' UCLA Law Review 38 (1991): 637.
107. Friedman, Infinite, supra note 37;
108. Friedman, Eclectic, supra note 39; Nance, supra note 30.
109. Two possible objections to an account based on the relative plausibility of explanations have already been discussed. These are that there may be too many or too few explanations. But as explained above, an explanation-based approach can adequately respond to these situations. Wesley Salmon has raised Bayesian challenges to IBE in the domain of science, but he recognized a legitimate domain for IBE in explaining everyday affairs such as explaining intentional behavior, where IBE does quite well and where the situation is too complex, ambiguous, and based on subtle linguistic cues to be reduced to a formalized process. (In other words, the kinds of situations that give rise to litigation.) See Wesley Salmon, 'Explanation and Confirmation: A Bayesian Critique of Inference to the Best Explanation,' and 'Reflections of a Bashful Bayesian: A Reply to Peter Lipton,' in Explanation: Theoretical Approaches and Explanations 61-91, 121-36 (G. Hon and S.S. Rakover eds.) (Norwell, MA and Dordrecht: Kluwer Academic Publishers, 2001).
110. Friedman, Eclectic, supra note 39 at 2045.
111. Nance, supra note 30 at 1568.
112. See Nance supra note 30 at 1575-84;
113. Friedman, Eclectic supra note 39;
114. Friedman, Infinite supra note 37.
115. Friedman, Infinite 37 at 93-94 n. 40.
116. Nance, supra note 30 at 1575-84;
117. Friedman, Eclectic supra note 39;
118. Friedman, Infinite supra note 37 at 93-94 n. 40.
119. See Friedman, Eclectic supra note 39;
120. Friedman, Infinite supra note 37 at 93-94 n. 40. Specifying more detailed stories means more stories with lower individual probabilities; less detailed stories means fewer stories with higher individual probabilities but the aggregate for both would remain the same.
121. Nance, supra note 30 at 1578.
122. See id. ("this 'third story' possibility is the most serious problem for the relative plausibility theory, one that needs to be addressed adequately before it is embraced.").
123. N.B., they would have relied on explanatory criteria to get these numbers.
124. Nance, supra note 30 at 1580-81.
125. Nance and Friedman both contend that relative plausibility must collapse into a Bayesian approach if it concludes that the defendant should win. Nance, supra note 30;
126. Friedman, Eclectic, supra note 39. We respectfully disagree. First, there is nothing problematic about a disjunctive explanation; and second, nothing about a disjunctive explanation suggests that it necessarily must be Bayesian. It, of course, might be, but this is not a challenge to relative plausibility. This seems to be similar to the fact that inductive inferences can often be re-cast as either abductive or enumerative. See infra note 15.
127. We thus concur with Friedman's statement that, "the observer's understanding of the world and of the issues at stake will usually provide considerable guidance on how to aggregate possibilities in a sensible way." Friedman, Eclectic, supra note 39 at 2042. As we discussed above, both the observer's understanding of the world and the issues and stake help to guide inferences based on explanatory criteria.
128. This criticism of Friedman's of explanation-based accounts of juridical proof, see supra note 37 and accompanying text, like most of the criticisms from the probabilists, is equally applicable to their own account. If the law requires that the precise time (whatever that might mean given the infinite series between any two points on a time line) be established, under a probabilist's account it would have to be established by a preponderance of the evidence (in civil cases) or (beyond reasonable doubt) in criminal cases. How that could be done is a mystery. We are fairly certain that the probabilists would view this as an unfair criticism of their approach for precisely the same reasons that we do: i.e., it involves a misconstruction of the law and amounts to criticizing an approach for being unable to do what the law does not demand and that no other approach could accomplish either.
129. Therefore, Nance, supra note 30, is incorrect to assert that if relative plausibility is allowed to aggregate potential explanations, it must aggregate all of them, which leads to a probability of 1. This ignores the contrastive nature of explanations.
130. Without reducing to the Bayesian account.
131. In a context, however, where the appropriate contrast is "what caused the heartburn: cheeseburgers or Thai food?" this is no longer a good explanation (and an inappropriate aggregation).
132. Nance, supra note 30 at 1609.
133. See Lipton, supra note 1 at 103-20.
134. On the paradox see Ronald J. Allen and Sarah A. Jehl, 'Burdens of Proof in Civil Cases: Algorithms vs. Explanations,' Mich. St. DCL L. Rev. (2003): 893;
135. Alex Stein, 'Of Two Wrongs that Make a Right: Two Paradoxes of the Law of Evidence and their Combined Justification,' Texas Law Review 79 (2001) 1199;
136. Saul Levmore, 'Conjunction and Aggregation,' Michigan Law Review (2001): 723.
137. This assumes that the elements are independent from one another (viz., that the likelihood on one element being true does not affect the likelihood of any other element being true), which often will not be the case in the law. Dependence makes things even more complicated for the Bayesian approach because one needs to know not only the probabilities of each element but how they interact. We put this additional problem for the Bayesian approach to the side. But we note that the explanation-based approach avoids this problem because the accepted explanation must incorporate each of the elements.
138. Friedman, Infinite, supra note 37 at 98.
139. Allen and Jehl, supra note 95.
140. Id. at 929-36. The preponderance rule is based on the assumptions that the parties should be treated roughly equally (and hence errors distributed roughly evenly among them) and that the evidence will generally favor the side that deserves to win (hence minimizing the total number of errors). For more detailed discussions of standards of proof with regard to error distribution see Laudan, supra note 48 at 63-88;
141. Richard S. Bell, 'Decision Theory and Due Process: A Critique of the Supreme Court's Law Making for Burdens of Proof,' Journal of Criminal Law and Criminology 78 (1987): 557.
142. See Friedman, Infinite, supra note 37 at 97 n. 48 (offering an alternative instruction that requires that the conjunction of elements be more likely than not true);
143. See also Dale A. Nance, 'A Comment of the Supposed Paradoxes of a Mathematical Interpretation of the Logic of Trials,' 66 Boston University Law Review 66 (1986): 947 (suggesting that each element individually plus their conjunction must meet the standard).
144. Friedman, Infinite, supra note 37 at 98 n. 50.
145. Even in such cases, instructing the jury to treat them as separate elements may increase the risk of error.
146. The same applies, with the necessary adjustments, for criminal cases and "clear and convincing" cases.
147. Juries are not explicitly instructed to do this, but it is plausible to suppose that they do so because explanatory criteria are used to infer holistic narratives of events before receiving jury instructions. See Pennington and Hastie, supra note 3;
148. Shari Seidman Diamond, Mary R. Rose, and Beth Murphy, 'Revisiting the Unanimity Requirement: The Behavior of the Non-Unanimous Civil Jury,' 100 Northwestern University Law Review (2006): 201, 212 ("The deliberations of these 50 cases revealed that jurors actively engaged in debate as they discussed the evidence and arrived at their verdicts. Consistent with the widely accepted 'story model,' the jurors attempted to construct plausible accounts of the events that led to the plaintiff's suit. They evaluated competing accounts and considered alternative explanations for outcomes.").
149. Schum, supra note 3 at 1655, has criticized IBE on a similar ground: "If we say we have the 'best' explanation . . . we must also be assured we have canvassed all possibilities." The legal system, however, addresses this concern by allowing parties to present the explanations they believe to be most favorable. There are an infinite number of possible explanations; it would, of course, be nonsensical to construct a decision procedure requiring that they
150. Friedman has thus posed the following challenge: I cannot recall the Bayesioskeptics ever offering any criticism about particular uses of probabilistic methods as a tool for analyzing evidentiary questions; the challenges always seem to be at the general level, concerning the value of the enterprise itself or the overall standard of persuasion. Richard D. Friedman, 'Answering the Bayesioskeptical Challenge,' International Journal of Evidence and Proof 1 (1997): 276, 290 (1997). We offer criticism of the use of such methods at the micro-level in Allen and Pardo, supra note 4.
151. Fed. R. Evid. 403.
152. For examples see the articles cited in note 2.
153. Nance and Morris, supra note 2.
154. Davis and Follette, supra note 2.
155. Finkelstein and Levin, supra note 2.
156. As mentioned earlier, the probative value could be defined as either the likelihood ratio, see Nance and Morris, supra note 2, or as the difference between the prior probability without the evidence and the posterior probability with it, Davis and Follette, supra note 2.
157. See Nance and Morris, supra note 2, who construct a "Bayesian norm" to measure juror performance in assessing DNA evidence. Jurors are considered correct or rational to the extent their assessments match the "Bayesian norm" and incorrect to the extent to which they deviate from it. We critique these studies in Allen and Pardo, supra note 4.
158. Amos Tversky and Daniel Kahneman, 'Evidential Impact of Base Rates,' in Judgement Under Uncertainty: Heuristics and Biases (Kahneman, Paul Slovic and Tversky eds.) (Cambridge University Press, 1982).
159. Id. at 157. The probability that the cab was blue given the testimony is calculated via Bayes' Theorem as .80 × .15/(.80 × .15) + (.20 × .85) = .12/.12 + .17 = .41.
160. Id.
161. Id. at 156 ("From a normative standpoint, however, the causal and the incidental base rates in these examples should have roughly comparable effects.").
162. Id. at 158-59. They continue: "This inference accounts for the differential base rates of accidents and implies that any Green cab is more likely to be involved in an accident that any Blue cab") Id. at 159.
163. The reference-class problem and its limitations for mathematical models of evidence is discussed in more detail in Allen and Pardo, supra note 4.
164. See Allen and Pardo, supra note 4.
165. Inferential interests pick out which variables are relevant.
166. Indeed, this may best explain why the median in the second scenario was .60 (rather than, say, .85 or .41).
167. The authors' ultimate conclusion is the following: "The major conclusion of this research is that the use or neglect of consensus information in individual prediction depends critically on the interpretation of that information." Tversky and Kahneman, supra note 114 at 160. To that we would add that it will be explanatory considerations driving that interpretive process. And rightly so.
168. The best way to do so appears to be to allow for subjective assessments based on explanatory criteria. For an example see Friedman, supra note 75.
169. For examples see Allen and Pardo, supra note 4.
170. See Pardo, supra note 54.
171. See Allen, supra note 31 at 630-31.
172. See Fed. R. Evid. 401, 403. The additional federal rules regulating relevance do so for the most part for policy reasons beyond the logical relevance or probative value of the evidence. See Fed. R. Evid. 404-15.
173. Id. at 632.
174. Such devices may be justified on grounds of both efficiency and to ensure the reliability or social plausibility of judgments. See id. at 632.
175. The assessment of all testimony requires an abductive inference about what best explains why the witness is making the assertions. See Thagard, supra note 2;
176. Jonathan E. Aldler, 'Testimony, Trust, Knowing,' Journal of Philosophy 71 (1994): 264, 274-75.
177. Even though DNA evidence may be presented as a "random match" probability, this number is not its probative value. See Allen and Pardo, supra note 4 (discussing Nance and Morris, supra note 2). Fact-finders must still make inferential judgment about what explains the result, taking account of such possibilities as whether lab error occurred, whether the evidence could have been planted, and whether other suspects may share similar DNA (with mitochondrial DNA, for example, whether the suspects share the same mother).
178. See, for example, the Blue-cab example discussed above.
179. That judges ought, and do, engage in more contextual judgments further vindicates the explanation-based account as descriptively accurate. See Old Chief v. United States, 519 U.S. 172 (1997);
180. Anderson v. Griffin, 397 F.3d 515 (7th Cir. 2005) ("Events that have a very low antecedent probability of occurring do sometimes occur ... and if in a particular case all the alternatives are ruled out, we can be confident that the case presents one of those instances in which the rare event did occur.") And, as the Blue-cab example illustrated, this practice is also epistemically sound.
181. See Anderson v. Liberty Lobby, Inc. 477 U.S. 242, 250-57 (1986).
182. See, e.g., Arthur R. Miller, 'The Pretrial Rush to Judgment: Are the "Litigation Explosion," "Liability Crisis," and Efficiency Cliches Eroding our Day in Court and Jury Trial Commitments?,' N.Y.U. Law Review 78 (2003): 982;
183. James Joseph Duane, 'The Four Greatest Myths about Summary Judgment,' Washington & Lee Law Review 52 (1996): 1523, 1554-62.
184. The Supreme Court employed an explanatory approach in a summary-judgment case in City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 437 (2002) ("Neither the Court of Appeals, nor the respondents, nor the dissent provides any reason to question the city's theory. In particular, they do not offer a competing theory, let alone data, that explains why the elevated crime rates in neighborhoods with a concentration of adult establishments can be attributed entirely to the presence of walls between, and separate entrances to, each individual adult operation.") In the antitrust context, the Court's opinion in Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587 (1986), created some confusion by its phrasing: "if the factual context renders respondents' claim implausible-if the claim is one that simply makes no economic sense-respondents must come forward with more persuasive evidence to support their claim than would otherwise be necessary." But as the phase "no economic sense" suggests, the standard is not whether the court thinks an explanation is implausible, it is whether any reasonable jury could find it to be plausible. The Court later clarified that the standard is the latter. See Eastman Kodak Co. v. Image Tech. Serv. Inc., 504 U.S. 451, 468-69 (1992).
185. Jackson v. Virginia, 443 U.S. 307, 324 (1979) (articulating standard as whether "[n]o rational trier of fact could have found proof of guilt beyond a reasonable doubt.") Defendants may also file "motions for acquittal" at trial based on the same sufficiency standard. See Fed. R. Crim. P. 29.
186. For examples employing these considerations, see United States v. Beard, 354 F.3d 691, 693 (7th Cir. 2004) ("Relative to the alternatives, the government's case was more powerful than it would have seemed in the abstract."); United States v. Newell, 239 F.3d 917, 920 (7th Cir. 2001). When the defendant does not offer a particular explanation of the evidence, but instead attempts only to poke holes in the prosecution's theory, the Jackson standard may still be met by showing that the prosecution's theory is implausible. In such a case, the reasonable inference is not that there is a plausible explanation which implies innocence, but rather that the prosecution has not met its burden of providing a plausible explanation.
187. Laudan, supra note 45.
188. Federal Civil Jury Instructions of the Seventh Circuit 34 (available at www.ca7.uscourts.gov). The instructions define the "clear and convincing" standard as "you are convinced that it is highly probable that it is true." Id. at 35.
189. Some states and federal circuits either require judges not to define the standard or do not require a definition. The Supreme Court has concluded that a definition is not required. Victor v. Nebraska, 511 U.S. 1, 5 (1994) ("the Constitution neither prohibits trial courts from defining reasonable doubt nor requires them to do so as a matter of course.").
190. In Victor, for example, the Court discussed whether definitions of the standard that use the phrase "to a moral certainty" were problematic.
191. Laudan, supra note 48 at 51-62.