In the late 1940s, doctors began prescribing DES (diethylstilbestrol, a synthetic estrogen developed in England in the 1930s) to prevent miscarriages. Years later it was discovered that DES could injure the reproductive system of the daughters of women who had taken it. Penelope Krist claims to be one of those injured daughters. She brought this tort suit in 1987 against Eli Lilly and Company, which along with a number of other drug companies had in 1947, the year before Penelope Krist's mother became pregnant, begun making and selling DES for the prevention of miscarriages. Under Wisconsin law, which all agree governs the substantive issues in this diversity suit, all that Penelope Krist had to prove in order to recover damages from Lilly was that her mother had taken DES when pregnant with Penelope (which is conceded for purposes of summary judgment); that the DES caused the injury for which she sues (also conceded for those purposes); and that Lilly had at the time when Penelope's mother was pregnant with her "produced or marketed the type (e.g., color, shape, markings, size, or other identifiable characteristics) of DES taken by the plaintiff's mother." Collins v. Eli Lilly Co., 116 Wis. 2d 166, 193, 342 N.W.2d 37, 50 (1984) (emphasis in original). It is on this last element that Lilly took its stand and persuaded the district court, on the eve of trial, to grant summary judgment and dismiss the suit.

The medical records of Penelope Krist's mother from 1948 do not indicate the brand of DES she took, but the mother (who is 68 years old) testified at her deposition in 1987 about the type of DES she took. She recalled taking, several times a day for at least six straight weeks, small, shiny, smooth, coated, red pills shaped "much like an M & M candy" (that is, convex rather than flat like an aspirin), only smaller--approximately half the size of an aspirin. She testified that in 1952 she had had a total hysterectomy, after which her doctor had prescribed DES to replace the estrogen lost by the removal of her ovaries, and that she had told her husband that the pills she received in 1952 were the same kind that she had taken during her pregnancy in 1948. When this suit was filed she was still taking the pills prescribed back in 1952--and the pills she was taking were made by Lilly. If they were the same pills she had been taking not only in 1952 but also in 1948, then probably those pills were made by Lilly too; it is on this theory that the plaintiff's counsel named Lilly as the defendant.

After the mother had signed her deposition but before the statute of limitations for suing other manufacturers of DES had expired, Lilly's counsel informed the plaintiff's counsel that Lilly had not manufactured red DES pills in 1948. Lilly later submitted evidence to that effect (and also to the effect that it had not manufactured coated DES pills in 1948, as she had also testified). The plaintiff does not contest this evidence, yet she failed either to change her theory of the case (as she might have done, we shall see) or to join any additional defendants; her counsel told us that it would have been beyond the plaintiff's financial means to do so. It is now too late for her to join additional defendants. Although Lilly brought additional drug manufacturers into the case as third-party defendants, to share the burden of liability should the plaintiff win a judgment against it, Collins v. Eli Lilly Co., supra, 116 Wis.2d at 195, 342 N.W.2d at 51, the plaintiff concedes that she cannot go after them directly: Lilly's decision to sue them did not waive their statute of limitations defense against the plaintiff's claim.

The district court granted Lilly's renewed motion for summary judgment when, on the morning of trial, the plaintiff's counsel stated that he intended to call only one witness, the plaintiff's mother, and added--with admirable candor--that she would testify that the pills she had taken while pregnant with Penelope were small, round, red pills, just as described in her deposition.

The deposition testimony concerning the color and other properties of the pills that Penelope's mother took during the pregnancy was given without hesitation, qualification, equivocation, uncertainty, misgiving. She expressed no doubt about the accuracy of her recollection; so far as we can glean from the transcript of the deposition, she testified with serene confidence. There are no contradictions, either within the deposition itself or between the deposition and any other evidence in the record, concerning the color of the pills. Lilly had begun manufacturing red DES pills after 1948 but before the prescription issued to the plaintiff's mother in 1952; and while if no drug company had been manufacturing red DES pills in 1948 it would be plain that the mother's recollection was faulty, no such evidence was presented. (On the contrary, the record contains a letter from McNeil Laboratories indicating that it manufactured red DES pills in the period when Penelope Krist's mother was pregnant with her, although it is uncertain whether the McNeil pill was distributed in the area where she lived.) If the mother's testimony is believed--and the plaintiff's counsel has suggested no reason not to believe it except that it concerned events almost forty years in the past--then it is plain that Lilly did not cause the injury of which Penelope Krist complains.

In resisting summary judgment, the plaintiff's counsel argues that he would have asked the jury to disbelieve the mother's testimony about the color (and coating) of the pills and that the jury might have done so. We are to evaluate the argument by asking whether, if the evidence at trial turned out to be the same as the pretrial evidence--and thus if the mother's testimony would be the same at trial as the testimony in her deposition--the judge would be required to grant a directed verdict for the defendant. If so, the defendant was entitled to summary judgment. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 250-52, 106 S. Ct. 2505, 2510-11, 91 L. Ed. 2d 202 (1986). The test for directed verdict, in turn, is whether a rational jury could have brought in a verdict for the nonmovant--the plaintiff in this case. Kozlowski v. John E. Smith's Sons Co., 87 Wis. 2d 882, 898, 275 N.W.2d 915, 922 (1979). (We use the Wisconsin test because in a diversity case the test for directed verdict or judgment notwithstanding the verdict is supplied by state law. Horton v. Miller Chemical Co., 776 F.2d 1351, 1355 (7th Cir.1985).) So, to state the question before us as precisely as possible, it is whether a rational jury deciding the case on the identical record that was before the judge when Lilly moved for directed verdict could have brought in a verdict for the plaintiff.

The jury would be confronted with a case in which a witness friendly to the plaintiff (this is an understatement) testified positively that the pills she took could not have been made by the defendant and in which irrefragable evidence established that those pills could have been made by someone other than the defendant. The only evidence in the record, therefore, would be evidence that Lilly was not the source of the plaintiff's injury. Lapse of time can distort recollection, but it does not always have that effect and the plaintiff has made no offer to prove that it usually does. The confidence and apparent completeness of the mother's recollection do not assure its accuracy, but neither do they provide leverage for an argument that it is inaccurate.

A party does not vouch for the truth of every statement of the witnesses he calls, even of the friendly witnesses. Fed.R.Evid. 608 and Note of Advisory Committee on Proposed Rule; Guenther v. Armstrong Rubber Co., 406 F.2d 1315, 1317-18 (3d Cir.1969). But with an exception that the plaintiff's counsel might have tried to fit this case within--but did not try--Wisconsin law requires the plaintiff in a DES case to prove that the defendant made the type of pill that the plaintiff's mother took while pregnant with the plaintiff, and Penelope Krist's mother is her only witness. How is the plaintiff to carry her burden of proof if the jury disbelieves her only witness? In theory the jury might disbelieve the mother's testimony about the color and coating of the pills but believe her testimony about their other attributes. But what sense would that make? The deposition indicates no greater uncertainty about color or coating than about other features of the pill--on the contrary, the color is the feature most stressed in her testimony. How could the jury rationally conclude that she had gotten the color and coating wrong but the size, shape, and other features connecting the pill to Lilly right?

There may be answers to these questions, but answers that come out of a scholarly literature of which the plaintiff's counsel appears to be unaware and which he in any event made no attempt to present through the affidavit of an expert who might later testify at the trial, as in such cases as United States v. Smith, 736 F.2d 1103 (6th Cir.1984) (per curiam), and United States v. Moore, 786 F.2d 1308, 1311-13 (5th Cir.1986). An important body of psychological research undermines the lay intuition that confident memories of salient experiences (such as taking a red pill for many weeks during pregnancy in an effort to prevent a miscarriage) are accurate and do not fade with time unless a person's memory has some pathological impairment. Much of this evidence can be found in Credibility Assessment (Yuille ed. 1989); Eyewitness Testimony: Psychological Perspectives (Wells & Loftus eds. 1984); Evaluating Witness Evidence (Lloyd-Bostock & Clifford eds. 1983). A leading scholar in this field is Elizabeth F. Loftus, author of papers in each of the three volumes we have cited and co-author of one of the volumes.

The basic problem about testimony from memory is that most of our recollections are not verifiable. The only warrant for them is our certitude, and certitude is not a reliable test of certainty. Many people are certain that God exists. Many are certain that He does not exist. The believer and the nonbeliever are equally certain, but they cannot both be correct. Similarly, the mere fact that we remember something with great confidence is not a powerful warrant for thinking it true. It therefore becomes an empirical question whether and in what circumstances memory is accurate. Cognitive psychologists such as Loftus have tried to answer this question. The answers are controversial. Zaragoza & McCloskey, Misleading Post-event Information and the Memory Impairment Hypothesis: Comment on Belli and Reply to Tversky and Tuchin, 118 J. Experimental Psych. 92 (1989). They are based for the most part on experiments with college students, and as with much experimentation in the social sciences it is uncertain how well the experimental results generalize to "real world" situations. But although the answers certainly are not definitive, they are suggestive. The basic findings are: accuracy of recollection decreases at a geometric rather than arithmetic rate (so passage of time has a highly distorting effect on recollection); accuracy of recollection is not highly correlated with the recollector's confidence; and memory is highly suggestible--people are easily "reminded" of events that never happened, and having been "reminded" may thereafter hold the false recollection as tenaciously as they would a true one.

All three of these findings could have been used in this case--how effectively we need not decide. The plaintiff's counsel wanted the jury to pick and choose among the mother's recollections of the features of the pills--to believe some and disbelieve others. It is difficult to see how the psychological research to which we have referred would have assisted counsel in this delicate operation. But it might have been used in another way: with the aid of an affidavit by an expert witness, the plaintiff's counsel might have changed the case from one in which the plaintiff can (and is therefore required to) allege and prove what type of DES the mother took to one in which the plaintiff cannot allege or prove this, in which event Wisconsin law allows her to get by with alleging and proving merely that the defendant produced or marketed DES for use in preventing miscarriage during the period in which the plaintiff's mother was pregnant. Collins v. Eli Lilly Co., supra, 116 Wis.2d at 143-44, 342 N.W.2d at 50.

The plaintiff's mother was testifying to events that happened forty years ago. Her confidence about those events had less significance than lay intuition would assign to such confidence of recollection. Her recollection that she had taken red pills in 1948 may have been the result of suggestion. She knew that the pills she was taking in 1987 were red DES pills made by Lilly. She knew that she had been taking the identical pills, made by the same company, for at least 25 years. She knew that if she was taking them in 1948 her daughter had a case against Lilly. She did not know (initially) that Lilly had not been making red DES pills in 1948. In these circumstances it was certainly possible for her to "recollect" that the pills she took in 1948 were red even if they actually were a different color. And once her memory was jogged she was likely to persist in her recollection even after she learned that she was not doing her daughter any good by remembering the pills as red. She is an honest person. She refused to change her testimony from what she actually, though quite possibly falsely, believed.

Since as we said a party is not required to vouch for everything his witness says, the plaintiff could have argued that her mother correctly recalled having taken DES but that her recollections concerning the type of DES she took were untrustworthy, thus making this a case where the plaintiff cannot prove what type of DES the mother took. Of course Lilly would be free to call the plaintiff's mother as its witness and try to prove from her testimony and McNeil's letter that Lilly did not manufacture the pills she took back in 1948. But the plaintiff could try in turn to counter the damaging parts of the mother's testimony with testimony by an expert witness about the vagaries of memory. It would be for the jury to decide whether the mother's recollection about the color of the pills was correct. The case therefore could not be decided on directed verdict, which means that it could not be decided in advance of trial, on summary judgment.

The plaintiff's counsel did not follow this route. He did not plan to present (presumably through an expert witness) grounds for disbelieving the testimony of the plaintiff's mother, which exonerated Lilly; he planned to make the mother the plaintiff's only witness. He intended to present the case not as one in which the plaintiff did not know what type of DES her mother had taken, but as one in which the plaintiff would prove that her mother had taken DES pills made by Lilly--nonred DES pills made by Lilly.

We do not want to be too hard on the plaintiff's counsel. In forgoing the expert-witness route he may have been concerned with language in opinions of this court that could be thought dismissive of expert testimony in the field of perception and memory. In United States v. Watson, 587 F.2d 365, 369 (7th Cir.1978), speaking of psychological studies of identification evidence in which the witness and the person identified are of different races, we said that "work in that field still remains inadequate to justify its admission into evidence." That was twelve years ago, and dealt with an esoteric topic as to which the defendants' own expert acknowledged that prior work was generally considered inadequate. Shortly before the oral argument in this case, another panel of this court remarked, concerning the exclusion of expert testimony on (among other topics) cross-racial identification, as in Watson, "such expert testimony will not aid the jury because it addresses an issue of which the jury already generally is aware, and it will not contribute to their understanding of the particular dispute." United States v. Hudson, 884 F.2d 1016, 1024 (7th Cir.1989). But this was in the context of upholding the district judge's decision to exclude the testimony. The question was whether the judge had abused his discretion, and the panel held he had not. This was not a holding that psychological evidence is never admissible on matters of perception and memory.

Although relevant evidence is presumptively admissible, Fed.R.Evid. 402, and the rule on expert testimony notably liberal, Fed.R.Evid. 702, a district judge has broad discretion to exclude relevant evidence that is confusing or redundant. Fed.R.Evid. 403. Not only is the cost of litigation raised by increasing the amount of evidence at trial, but the accuracy of litigation as a method for finding the truth is impaired when the additional evidence does more to distract or confuse the jury than to enlighten it. Often it does. Information is not effortlessly absorbed, especially information of a technical character. Certainly in routine cases the trial judge is not required to allow wide-ranging inquiry into the mysteries of human perception and recollection. But few cases involve recollections from forty years earlier. Such cases are not routine, and psychological evidence may be helpful to the judges and jurors required to decide them. "Perceptual psychology (a part of experimental psychology) is not 'junk science.' " Carroll v. Otis Elevator Co., 896 F.2d 210, 215 (7th Cir.1990) (concurring opinion). Carroll upheld the district judge's decision to allow a psychologist to testify, rejecting the objection that "the subjects upon which he testified"--the attractiveness of brightly colored objects to small children--"were within the ken of the average juror." Similarly in the present case the district judge might well have allowed expert psychological evidence on the vagaries of confident recollections of events lying decades in the past. But, again, the plaintiff's counsel did not want to take that route.

There is a tension in our decisions. Hudson and Watson evince a more skeptical view of expert evidence on perception than Carroll does. We need not try to reconcile the tension here. It may have more to do with the particulars of the cases in question than with any deep difference of opinion on the utilization of science in litigation; it may therefore be illusory. An irony of the present case is that the plaintiff's counsel might have been able to turn Hudsonian skepticism to his own advantage. If testimony on the psychology of perception and memory would merely be addressing "an issue of which the jury already generally is aware," this implies that, even without the aid of an expert witness, the jury in this case (if the case had been allowed to go to a jury) might have tumbled to the unreliability of the mother's testimony despite the confidence and circumstantial detail with which that testimony was given and the salience of the fact recalled. The jury might, in short, have disbelieved the mother on its own, and if it could have done so rationally, a directed verdict (and therefore summary judgment) would be improper. But remember that the plaintiff's counsel did not want the jury just to disbelieve the mother. If he had wanted that, he would not have called her as a witness. He wanted the jury to believe her description of every feature of the pills except their color and coating. Maybe a psychologist could have guided the jury to such a conclusion (although this seems unlikely), but without such guidance a rational jury could not have reached it. Indeed, it would have been given no chance to do so. Without an articulated theory of the mother's partial credibility, there would be no basis for concluding at the close of the plaintiff's case that the plaintiff had carried its burden of proof on the type of DES that her mother had taken; the judge would grant the defendant's motion for directed verdict.

There is latent in this discussion a further tension, however: between the orthodox view that treats the issue on summary judgment as whether the moving party would be entitled to directed verdict if the proceeding were a trial rather than a summary judgment proceeding, and the "realist" view that treats the issue on summary judgment as whether the nonmovant has a prayer of winning at trial. Mason v. Continental Illinois Nat'l Bank, 704 F.2d 361, 367 (7th Cir.1983). Although it is inconceivable that a real jury, unless consumed by bias, would conclude from the mother's testimony and the other evidence in the case that Lilly had manufactured the DES pills that she took when pregnant with the plaintiff, a superrational jury able to intuit the perceptual vagaries that Professor Loftus and her colleagues have discovered through patient research might so conclude, and perhaps therefore a directed verdict would not be granted--since the test for whether to grant a directed verdict is whether no rational jury could find for the nonmovant. But this reasoning is too finespun. The issue is not whether a jury composed of experts in perceptual psychology, presided over by a judge equally expert in that field, could find for the plaintiff in the absence of evidence in the usual sense. The issue is whether a normal jury could so find. It could not. (Compare the principle that there can be no finding of medical negligence in the absence of expert testimony to support it. Prosser and Keeton on the Law of Torts, Sec. 32, at p. 188 (5th ed. 1984).) Lacking scientific knowledge that an expert might have imparted to it but did not, no rational jury could bring in a verdict for the plaintiff, and therefore summary judgment was properly granted to Lilly.

Rationality is not a synonym for omniscience. If juries could be presumed omniscient, summary judgment could never be granted; the omniscient jury does not need evidence. In deciding what a rational jury could decide, we may not impute to it knowledge that it could have obtained only from expert testimony that no party was prepared to obtain.

The Wisconsin rule on proof of causation in DES cases (a rule that is not unique to Wisconsin, Martin v. Abbott Laboratories, 102 Wash.2d 581, 604, 689 P.2d 368, 382 (1984); Smith v. Eli Lilly & Co., 173 Ill.App.3d 1, 18-19, 122 Ill. Dec. 835, 846, 527 N.E.2d 333, 344, appeal granted, 123 Ill. 2d 567, 128 Ill. Dec. 900, 535 N.E.2d 411 (1988)) places a heavier burden on plaintiffs than the rule of the "market share" states. These states allow the DES plaintiff to recover damages upon showing that the defendant manufactured DES at the time of the mother's pregnancy--never mind what type of DES--and then allow the defendant through impleader to shift a part of the burden to the other companies that manufactured DES at that time. E.g., Hymowitz v. Eli Lilly & Co., 73 N.Y.2d 487, 511-12, 541 N.Y.S.2d 941, 950, 539 N.E.2d 1069, 1078 (1989). Wisconsin is a market share state only in cases where the plaintiff is unable to establish the type of DES that her mother took, and the plaintiff here claimed to be able to establish it. Perhaps the approach of the pure market share states is the better one, but Wisconsin has chosen differently (and even the Wisconsin rule is highly favorable to plaintiffs, compared to the traditional common law approach to proof of causation). We are bound by its choice, just as we are bound by the strategic decision of the plaintiff's counsel to try to prove what type of DES pills the plaintiff's mother took when pregnant with the plaintiff rather than to name additional defendants.

AFFIRMED.