By reason of an equally divided en banc court, the decision of the district court is AFFIRMED. The panel opinion was vacated by the grant of rehearing en banc.

would reverse for additional reasons:

I concur in Judge Clement’s fine opinion and would also highlight the troubling tension between this unprecedented multimillion dollar judgment against a major metropolitan District Attorney’s office and the policies that underlie the shield of absolute prosecutorial immunity. The Supreme Court ought to address whether holding governmental entities liable for Section 1983 violations is consistent with absolute prosecutorial immunity from such violations. Imbler v. Pachtman, 424 U.S. 409, 96 S.Ct. 984, 47 L.Ed.2d 128 (1976).

The Supreme Court recently issued a unanimous opinion affording absolute immunity from personal- Section 1983 liability to Los Angeles County’s chief prosecutors for failure to train or supervise their staff, or failure to establish appropriate systems in regard to the advocacy function of their office. Van de Kamp v. Goldstein, — U.S. -, 129 S.Ct. 855, 172 L.Ed.2d 706 (2009). Much as in this case, a plaintiff had been freed from custody after he discovered that important evidence had been withheld during his prosecution. The Court made a number of observations that *294are prescient of the circumstances leading to liability in this case. These bear repeating or paraphrasing with my editorial analogies to prosecutor’s offices.

1. The “public trust” in the prosecutor’s office would suffer were he to have in mind his own potential liability when making prosecutorial decisions. Van de Kamp, 129 S.Ct. at 860 (quoting Imbler). Likewise, public confidence will erode if the public believe a prosecutorial office is motivated by the impulse to cover itself when challenged by difficult prosecutions.

2. The frequency with which criminal defendants bring suits creates real fear about public perception as well as the independence of judgment exercised by prosecutors under the constant threat of lawsuits. Van de Kamp, id.

3. Such suits, whether against the prosecutor — or the office — , “ ‘often would require a virtual retrial of the criminal offense in a new forum, and the resolution of some technical issues by the lay jury.’ ” Van de Kamp, id. (quoting Imbler). See footnote 41 of Judge Clement’s opinion.

4. A prosecutor “ ‘inevitably makes many decisions that could engender color-able claims of constitutional deprivation.’ ” Van de Kamp, id. (quoting Imbler). See Judge Clement’s opinion at text adjoining footnote 53.

5. Defending against such claims, “ ‘often years after they were made, could impose unique and intolerable burdens upon a prosecutor [or office] responsible annually for hundreds of indictments and trials.’ ” Van de Kamp, id. (quoting Imbler). A crucial witness here had died, and other prosecutors could not recall this case as distinct from the hundreds or thousands they had handled.

6. The Court also said: “We do not see how ... differences in the pattern of liability among a group of prosecutors in a single office \i.e. distinguishing between the supervisors and the line prosecutors] could alleviate Imbler’s basic fear, namely, that the threat of damages liability would affect the way in which prosecutors carried out their basic court-related tasks.” Van de Kamp, 129 S.Ct. at 862. Moreover, “... ‘it is the interest in protecting the proper functioning of the office, rather than the interest in protecting its occupant, that is of primary importance.’ (internal citation omitted).” Id. Authorizing Section 1983 liability against the office creates the same stress on the proper functioning of the office.

7. With regard to liability for supervisory actions related to the trial process, the Court held that “a suit charging that a supervisor made a mistake directly related to a particular trial, on the one hand, and a suit charging that a supervisor trained and supervised inadequately, on the other, would seem very much alike.” Van de Kamp, 129 S.Ct. at 863.

8. “It will often prove difficult to draw a line between general office supervision or training and specific supervision or training related to a particular case.” Van de Kamp, id. “To permit claims based upon the former is inevitably to permit the bringing of claims that include the latter.” Id. In this case, the jury was permitted to infer Section 1983 deliberate indifference and causation based on a single incident of withheld Brady evidence.

9. “[0]ne cannot easily distinguish, for immunity purposes, between claims based upon training or supervisory failures related to Giglio [at issue in Van de Kamp] and similar claims related to other constitutional matters (obligations under Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), for example). And that being so, every consideration that Imbler mentions militates in favor of immunity.” Id.

10. If the threat of damages liability for a trial error could lead a trial prosecu*295tor to take account of that risk when making trial-related decisions, so, too, could the threat of office liability for the same error affect the decisions of other prosecutors. Van de Kamp, 129 S.Ct. at 863. So, too, could the office policies become infected with risk aversion.

11. Because “better training or supervision might prevent most, if not all, prosecutorial errors at trial, permission to bring such a suit [in Van de Kamp] would encourage claims [by other criminal defendants], in effect claiming damages for (trial-related) training or supervisory failings.” Van de Kamp, id. Such suits could, “given the complexity of the constitutional issues,” “pose substantial danger of liability even to the honest prosecutor.” Id. (quoting Imbler.) Indeed, only four convictions of the New Orleans District Attorney’s office were overturned for Brady violations in the decade preceding Thompson’s conviction (Judge Clement’s opinion at footnotes 49-50), and none involved lab reports.

12. Practical anomalies result from the coexistence of absolute prosecutorial immunity with potential Monell liability of the prosecutor’s office. As the Court observed in Van de Kamp, id., “[s]mall prosecution offices where supervisors can personally participate in all of the cases would ... remain immune from [damage suits]; but large offices, making use of more general office-wide supervision and training, would not.”

13. “Most important, the ease with which a plaintiff could restyle a complaint charging a trial failure so that it becomes a complaint charging a failure of training or supervision would eviscerate Imbler.” Van de Kamp, id. This seems true whether the potential defendant is a supervisor, as in Van de Kamp, or the governmental office itself, as in this case.

The Court has not specifically excluded municipal Section 1983 liability for prosecutorial offices, nor has it ruled that they are vulnerable. Still, every reason advanced in Van de Kamp and Imbler for protecting the independence and integrity of prosecutors in trial-related actions and supervision suggests that holding a government entity liable in their stead for the same violations is simply untenable. The Court recognized, “as Chief Judge Hand pointed out [in Imbler], that sometimes such immunity deprives a plaintiff of compensation that he undoubtedly merits; but the impediments to the fair, efficient functioning of a prosecutorial office that liability could create lead us to find that Imbler must apply here.” Van de Kamp, 129 S.Ct. at 864. Today’s judgment raises issues that will continue to plague honest prosecutors’ offices.

Specially Concurring:

Ordinarily, when an en banc case results in a tie vote, we affirm the district court judgment without opinion. That is the way I would prefer it today. However, notwithstanding that there is no majority opinion, and that no opinion today will bind any court or future party in this circuit, each side has now written for publication, and judges are joining one or the other of the respective opinions. I join Judge Clement’s opinion because, as between the two, it shows the intellectual fortitude of meeting head-on, in a specific workmanlike manner, the truly difficult legal issues presented by this case.

We believe it imperative to explain why the result in this case should not encour*296age the extension of single incident municipal liability under Monell. John Thompson, the plaintiff-appellee, was convicted of murder and spent fourteen years on death row for a crime he did not commit because prosecutors failed to turn over a lab report in a related case. In this 42 U.S.C. § 1983 case, he sought compensation for the years spent in prison and on death row. The jury awarded Thompson $14 million against the Orleans Parish District Attorney in his official capacity. This appeal asks whether Harry Connick, the District Attorney, was deliberately indifferent to an obvious need to train his assistants on their obligations under Brady v. Maryland;1 it further asks whether a lack in Brady training was the moving force behind Thompson’s constitutional injury. The district court, and a panel of this court, held that the evidence was legally sufficient to support both of these claims. The panel opinion was vacated by our order for en banc rehearing.

Only under the most limited circumstances may a municipality be held liable for the individual constitutional torts of its employees. Considering the strict standards of culpability and causation applicable here, we conclude that the evidence supporting Thompson’s claim was legally inadequate to hold the District Attorney’s Office liable ' for this employee failure. Along similar lines, we also conclude that the jury instructions given on “deliberate indifference” were plainly erroneous.

FACTS

In 1985, a few weeks before his murder trial, John Thompson was tried and convicted of attempted armed robbery. Because of the attempted armed robbery conviction, Thompson decided not to testify in his own defense in his trial for the murder of Raymond T. Liuzza, Jr. Thompson was convicted of murder and sentenced to death.

Fourteen years later, in 1999, an investigator in Thompson’s habeas proceedings discovered that prosecutors had failed to turn over a crime lab report in the attempted armed robbery case. That lab report indicated that the perpetrator had type B blood. Because Thompson has type 0 blood, the attempted armed robbery conviction was vacated. In 2002, the Louisiana Fourth Circuit Court of Appeals granted post-conviction relief and reversed Thompson’s murder conviction, holding that the improper attempted armed robbery conviction had unconstitutionally deprived Thompson of his right to testify in his own defense at his murder trial.2 Thompson was retried for. Liuzza’s murder in 2003 and found not guilty.

After his release, Thompson brought suit alleging various claims against the District Attorney’s Office, Connick, James Williams, Eric Dubelier, and Eddie Jordan' — -the District Attorney in 2003 — in their official capacities; and Connick in his individual capacity (collectively, “Defendants”).3 The only claim that proceeded to trial was a claim under § 1983 for wrongful suppression of exculpatory evidence in violation of Brady v. Maryland. Thompson presented two theories of liability to the jury: (1) that the Brady violation was due to an unconstitutional official policy of the District Attorney’s Office, and (2) that the Brady violation was caused by Con-nick’s deliberate indifference to an obvious need to train, monitor, or supervise his *297prosecutors. The jury found that the Brady violation was not due to an official policy of the District Attorney’s Office, but was due to a failure to train. The jury awarded Thompson $14 million in damages. Defendants filed timely motions for judgment as a matter of law — before and after the verdict — as well as a motion to amend or alter the judgment and a motion for a new trial.4 The district court denied all of these motions, and Defendants appealed.

STANDARD OF REVIEW AND APPLICABLE LAW

We review the denial of a motion for judgment as a matter of law de novo.5 Judgment as a matter of law is appropriate if “the court finds that a reasonable jury would not have a legally sufficient evidentiary basis to find for the party on that issue.”6 Under this standard, we consider all of the evidence “in the light and with all reasonable inferences most favorable to the party opposed to the motion.”7 Substantial evidence — defined as “evidence of such quality and weight that reasonable and fair-minded men in the exercise of impartial judgment might reach different conclusions” — must be presented by the non-moving party, and a “mere scintilla” is insufficient.8 “[A] jury’s freedom to draw reasonable inferences does not extend so far as to allow the jury to draw an inference which amounts to mere speculation and conjecture.”9 In reviewing the record, we “give credence to the evidence favoring the nonmovant as well as that evidence supporting the moving party that is uncontradicted and unimpeached.”10

Because we are reviewing a Monell11 verdict against a government entity, our evidentiary review must take into account that § 1983 Monell liability is fundamentally different from an entity’s vicarious liability, predicated on respondeat superi- or, for its employees’ ordinary misconduct. Thus, when a plaintiff seeks to impose § 1983 liability on a municipality for its failure to train its employees, normal tort standards are replaced with heightened standards of culpability and causation.12 *298Liability will only attach if the municipality was deliberately indifferent to the constitutional rights of citizens — a showing of negligence or even gross negligence will not suffice.13 Errors of judgment do not alone prove deliberate indifference, nor is such heightened culpability established simply by showing that a municipality could have ordered more or different training or even misjudged whether training was necessary. Similarly, to satisfy causation, the plaintiff must demonstrate that the failure to train was the moving force behind the constitutional violation — “but for” causation is not enough.14 The Supreme Court, has repeatedly cautioned that if we neglect these stringent standards, we risk collapsing the distinction between vicarious liability and direct liability.15 Heightened standards also guard against the potentially “endless exercise of second-guessing municipal employee-training programs,” a task for which federal courts are ill suited.16

A. Heightened Culpability: Deliberate Indifference

Municipal liability attaches “where — and only where — a deliberate choice to follow a course of action is made from among various alternatives by city policymakers. Only where a failure to train reflects a ‘deliberate’ or ‘conscious’ choice by a municipality — a ‘policy’ as defined by our pri- or cases — can a city be liable for such a failure under § 1983.”17 Because a municipality rarely decides not to train its employees out of sheer contempt for citizens’ rights, deliberate indifference must be proven circumstantially. Deliberate indifference “generally requires” a showing that the policymaker was made aware of the training deficiencies by “at least a pattern” of similar deprivations.18 Absent a pattern, in certain unique circumstances, the plaintiff can establish liability based upon a single violation of constitutional rights. In such a case, a failure to train constitutes municipal policy only where the need for training was “so obvious” that a failure to do so would mean that the policymaker was deliberately indifferent to constitutional rights.19 A need for training is considered sufficiently obvious only when the deprivation of constitutional rights is a “highly predictable consequence” of the training deficiency.20 “Deliberate indifference” implies a sense of callousness21 and is “treated, as it is elsewhere in the law, as tantamount to intent, so that inaction by a policymaker deliberately indifferent to a substantial risk of *299harm is equivalent to the intentional action that [the] setting [of] policy presupposes.”22 The Supreme Court has given us a hypothetical example of such callousness: giving untrained police officers firearms and ordering them to arrest fleeing suspects. In such a situation, the deprivation of constitutional rights, i.e., the use of excessive force, is a highly predictable consequence of the lack of training, and therefore the need for such training is sufficiently obvious.23

Our court has considered single violation liability several times, and, with only one exception in some thirty years since Monell, has “consistently rejected application of the single incident exception.”24 The sole exception, Brown v. Bryan County, involved a failure to train a neophyte on the constitutional limits to the use of force.25 The Brown court focused on the decision by the sheriff to place his nephew, a completely untrained new deputy, “on the street to make arrests.”26 Within a matter of weeks, the deputy unconstitutionally injured a citizen he was trying to arrest. This court concluded, based on an extensive list of factors, that “the jury reasonably could have concluded that Sheriff Moore made a conscious decision not to train [his nephew], yet still allowed him to make arrests.”27 There was “unmistakable culpability and clearly connected causation” sufficient to hold the municipality liable.28

In Brown, the conscious decision was to not train a specific deputy, and the excessive use of force — which was the “highly predictable consequence” of failing to train the deputy — occurred soon after the officer went out on the streets. Several panels of this court, however, have reviewed cases where a decision not to train was made long before the alleged violation, and found that the lack of any similar violations indicates that a violation could not be the “highly predictable consequence” of failing to train.29 This approach reflects *300common sense: if there have been thousands of opportunities for municipal employees to violate citizens’ constitutional rights, and yet there have been no previous violations, then the need for training is simply not “so obvious.”30

B. Heightened Causation: Moving Force

To safeguard the boundaries established in Monell, the Supreme Court has made clear that in addition to a heightened standard of culpability, plaintiffs must meet a heightened standard of causation in order to hold a municipality liable under § 1983.31 A § 1983 plaintiff must prove that the municipal entity’s custom or policy — in this case the failure to train — was the “moving force” that caused the specific constitutional violation; stated differently, the plaintiff must establish a “direct causal link” between the municipal policy and the constitutional injury.32 We have said that the “connection must be more than a mere ‘but for’ coupling between cause and effeet.”33 The deficiency in training must be the actual cause of the constitutional violation.

Accordingly, the District Attorney must not be held liable simply because the culpable assistant district attorneys worked for him. “[A] municipality can be found liable under § 1983 only where the municipality itself causes the constitutional violation at issue.”34

That this heightened standard is vital to maintaining Monell’s prohibition against vicarious liability in § 1983 cases has been clearly expressed by the Supreme Court:

Where a court fails to adhere to rigorous requirements of ... causation, municipal liability collapses into respondeat superi- or liability. As we recognized in Monell and have repeatedly reaffirmed, Congress did not intend municipalities to be held liable unless deliberate action attributable to the municipality directly caused a deprivation of federal rights.

*301A failure to apply stringent culpability and causation requirements raises serious federalism concerns ....35

And in City of Canton, the Supreme Court further said:

To adopt lesser standards of fault and causation would open municipalities to unprecedented liability under § 1983. In virtually every instance where a person has had his or her constitutional rights violated by a city employee, a § 1983 plaintiff will be able to point to something the city “could have done” to prevent the unfortunate incident. See Oklahoma City v. Tuttle, 471 U.S. [808] at 823[, 105 S.Ct. 2427, 85 L.Ed.2d 791 (1985)] (opinion of Rehnquist, J.). Thus, permitting cases against cities for their “failure to train” employees to go forward under § 1983 on a lesser standard of fault would result in de facto respondeat superior liability on municipalities — a result we rejected in Monell, 436 U.S. at 693— 694[, 98 S.Ct. 2018], It would also engage the federal courts in an endless exercise of second-guessing municipal employee-training programs. This is an exercise we believe the federal courts are ill suited to undertake, as well as one that would implicate serious questions of federalism.36

This exacting standard is even more crucial in cases where the municipal policy is said to be a failure to train. Because such a “policy,” standing alone, implicates no constitutional violation, the “causal connection between municipal policy and the deprivation of constitutional rights becomes more difficult to discern.”37 Thus, a careful analysis is required to determine “whether a jury reasonably could conclude that the [municipality’s] conduct was the moving force in bringing about the constitutional violation.”38

To summarize, the requirements for imposing liability upon a municipality for the individual acts of its employees are demanding. Relaxing these heightened requirements would cause significant harm to the interests underlying this demanding evidentiary principle: “adopting] lesser standards of fault and causation would open municipalities to unprecedented liability,” “would result in de facto respondeat superior liability,” and would “engage the federal courts in an endless exercise of second-guessing municipal employee-training programs.”39 Therefore, we can hold a municipality liable only where the evidence demonstrates “unmistakable culpability and clearly connected causation” for the unconstitutional conduct of an individual employee.40

DISCUSSION

A. Sufficiency of the Evidence — Culpability

The Brady violation here was a failure of one or more of the four assistant district attorneys involved with Thompson’s armed robbery prosecution to turn over the crime lab report to Thompson’s counsel.41 It is undisputed that the District *302Attorney’s Office did not provide formal in-house training regarding Brady.42 It is also undisputed that the assistant district attorneys were familiar with the general rule of Brady that evidence favorable to the accused must be disclosed to the defense.43 Thompson’s burden was to prove that Connick, the policymaker for the Orleans Parish District Attorney’s office, was deliberately indifferent to the need to train prosecutors in their Brady disclosure obligations.

As this circuit has recently recognized, “it is not enough for [Thompson] to show that the [District Attorney’s Office’s] training program is, in a general sense, wanting.”44 Instead, “the identified deficiency in a city’s training program must be closely related to the ultimate injury.”45 The plaintiff must prove an affirmative to the question: “Would the injury have been avoided had the employee been trained under a program that was not deficient in the identified respect?”46 Every prosecutor understood his general duty under Brady, so the identified deficiency was not a failure to train on this general duty, but was instead a failure to train on how to handle specific types of evidence such as the crime lab report at issue.

There was evidence that some prosecutors doubted whether Brady itself obligated the production of evidence that was not necessarily exculpatory.47 This confusion *303seems to have arisen because the report itself had the potential to be either exculpatory or inculpatory — depending on whether it matched Thompson’s blood type.48 Accepting that there was no training on the Brady obligations pertaining to potentially exculpatory crime lab reports, we must determine whether the need for that training was “so obvious” that a reasonable jury could find that Connick was “deliberately indifferent” to that need.

Thompson did not show any pattern of similar Brady violations, and instead relies exclusively on this single incident where prosecutors failed to disclose his crime lab report. In another case before this comb, we sustained the district court’s conclusion that twenty-five years of records involving this District Attorney’s Office (covering the time period of Thompson’s trial) revealed no pattern of Brady violations.49 Connick testified that the District Attorney’s Office handled tens of thousands of cases annually around this time, and that in the ten years prior to Thompson’s case, only four convictions were overturned based on Brady violations, none of which was similar to the violation at issue.50 So in only a minute number of cases were convictions overturned based on Brady violations, and there was not a single instance involving the failure to disclose a crime lab report or other scientific evidence. Connick was not alerted to a need for further Brady training — especially not this specific type of Brady training — -by previous violations at the District Attorney’s Office.51

Nor has Thompson been able to refer us to a single reported opinion, issued before this 1985 prosecution, from the Supreme Court, any of the circuit or district courts, or any state court that involved a similar Brady violation and thus might have alerted Connick to the need for Brady training in this area.52

Thompson instead points to the following as substantial evidence that the need for training in this area was “so obvious” that a failure to train constituted deliberate indifference. First, Thompson argues that Connick testified that he knew his prosecutors would frequently come into contact with Brady evidence. Second, *304many prosecutors testified that the law regarding Brady contains “gray areas.” Third, Thompson noted that several of the assistant district attorneys were only a few years out of law school. Thompson also points to intra-office discussions and opinions of various assistant district attorneys from 1999 and later about whether the lab report was evidence covered by Brady.

This type of evidence amounts to no more than general observations that would apply to any area of law and any number of district attorney’s offices throughout the country. All district attorneys know that Brady issues — along with many other areas of constitutional law — are routine matters that their assistants handle every day. Every district attorney knows that nearly all issues he deals with are shaded with “gray areas,” whether they concern Brady, search and seizure, Miranda, evidence of a defendant’s other crimes, expert witnesses, sentencing, or many more.53 Incorrect prosecutorial decisions in any of these areas may lead to later reversal of convictions. Nearly all district attorney’s offices employ prosecutors only a few years out of law school. That there were different opinions about Brady evidence, or any other issue that may be raised among lawyers, should surprise no one. All of this evidence involves generic generalizations— not the type of exacting evidence required to show that Connick and the District Attorney’s Office were deliberately indifferent to an obvious need to further train its professional prosecutors. To the extent that this evidence could be used to show that the municipality’s training was, in a general sense, wanting, similar evidence could also support a deliberate indifference finding against any prosecutor’s office for nearly any error that leads to a reversal of a conviction.

We cannot accept the argument that generalized failure to train evidence sustains a finding of official deliberate indifference. In Pineda v. City of Houston,54 Judge Higginbotham squarely rejected a plaintiffs argument that “ ‘because the City has admitted that specialized training is required for officers in such situations [specialized narcotics investigations], there is sufficient evidence that the training was inadequate.’ ... No butterfly will emerge from this hollow chrysalis of an argument.”55

Pineda relies on City of Canton, which displayed utmost caution toward generalized failure to train evidence:

Neither will it suffice to prove that an injury or accident could have been avoided if an officer had had better or more training, sufficient to equip him to avoid the particular injury-causing conduct. Such a claim could be made about almost any encounter resulting in injury, yet not condemn the adequacy of the program to enable officers to respond properly to the usual and recurring situations with which they must deal. And plainly, adequately trained officers occasionally make mistakes; the fact that they do says little about the training program or the legal basis for holding the city liable.56

Because this case concerns the actions of licensed attorneys who have independent professional obligations to know and uphold the law, there is even more reason than in City of Canton or Pineda not to rely on generalized statements about lack of training. Training is what differenti*305ates attorneys from average public employees. A public employer is entitled to assume that attorneys will abide by the standards of the profession, which include both ethical and practical requirements. Thus, prosecutors are personally responsible as professionals to know what Brady entails and when to perform legal research to understand the “gray areas.”57 To hold a public employer liable for failing to train professionals in their profession is an awkward theory. By analogy, it is highly unlikely that a municipality could be held liable for failing to train a doctor it employed in diagnostic nuances. Mere nostrums about training in Brady, a basic due process principle of criminal procedure, will not suffice.

While Thompson has failed to produce any specific evidence relating to the obvious need for training on the identified deficiency, there is sufficient evidence to show that the need for training in this area was in fact not “so obvious.” This situation — with scientific evidence providing a blood type or other indicator of the perpetrator’s identity — occurs every time a crime lab report is prepared and blood or other scientific evidence has not been taken from the defendant. And yet Thompson failed to show any similar failures to disclose crime lab reports from this District Attorney’s Office either before his trial or since. The alleged failure to train extended over a long period of time, during which hundreds or thousands of crime lab reports were prepared.

As a matter of probability, if violations were the “highly predictable consequence” of a failure to train, then we would expect to see more than just one violation in hundreds or thousands of cases. Thompson has, as a legal matter, failed to prove that his violation was the “highly predictable consequence” of failing to train prosecutors. This means that the need for training was not “so obvious,” and thus that Connick was not “deliberately indifferent” to Thompson’s constitutional rights.58

Indeed, three witnesses testified that the District Attorney’s Office had a policy of disclosing all crime lab reports. Dubelier testified that turning over crime lab reports was “standard operating procedure in the office.”59 Williams testified that “[y]ou have to turn over any scientific evidence that you have.” And Assistant District Attorney Val Solino testified that turning over crime lab reports was “the policy in the office.”60 While there was *306some confusion as to Brady obligations, every single witness who was asked stated that they would have disclosed the crime lab report had they known about it. This testimony was uncontradicted and unimpeached.61 Because the District Attorney’s Office had a policy of disclosing all crime lab reports, there was no need to train on the specifics of which reports would or would not be covered by Brady. Just as a municipal policymaker could not be found deliberately indifferent to citizens’ Brady rights if the policymaker established clear policies — such as an open-file policy — to protect those rights, Con-nick cannot be considered “deliberately indifferent” to a Brady violation based on a failure to disclose a crime lab report when his employees generally understood that they had to disclose exactly those types of reports.

For these reasons, under Monell, City of Canton, and Bryan County, the evidence in this record does not support the conclusion that Connick was deliberately indifferent to an obvious need for training. Consequently, the District Attorney cannot be held liable for the failure by his employees to disclose this crime lab report.

B. Sufficiency of the Evidence — Causation

Nor does the diffuse evidence of Brady misunderstanding among several assistant district attorneys satisfy the causation requirements relating to the violation at issue here. The specific question we ask is whether Connick’s failure to provide in-house training was the moving force behind the failure to turn over the lab report. Stated differently, the question is whether, under the teachings of City of Canton and Bryan County, unfamiliarity with Brady obligations with respect to this lab report was the actual cause — the moving force— of this constitutional violation. This standard of causation must be established by substantial evidence. If Thompson did not submit substantial evidence that the failure to produce the lab report was caused by confusion over Brady, the jury could not have reasonably concluded that the lack of training was the direct cause of Thompson’s injury, and judgment as a matter of law is required.

The record leaves many unanswered questions about the actual cause of this constitutional violation. It is, however, clear that four assistant district attorneys were involved in the failure to produce the lab report. They were Bruce Whittaker, James Williams, Gerry Deegan, and Eric Dubelier. The failure to produce the lab report lies with one or more of these assistant district attorneys. We turn now to examine what the record reflects as to these assistants.

*307Whittaker, as the armed robbery “screener,” was responsible for initially reviewing the police file on Thompson, deciding whether to prosecute, and assigning the case to the correct division. He “screened” Thompson’s file on February 25, 1985 — approximately five weeks after Thompson had been arrested. The police report indicated that evidence had been collected from the crime scene that possibly contained the perpetrator’s blood. Accordingly, Whittaker wrote “May wish to do blood tests,” on the screening action form.

After Whittaker screened the case, Dubelier, an experienced assistant district attorney, was assigned as lead prosecutor, and Deegan was assigned as the junior assistant on the case. Dubelier had no independent recollection of the armed robbery prosecution that had occurred twenty years earlier, but the record indicates that he and Deegan handled most of the pretrial work. Williams did not become involved with the case until March 11 when Dubelier asked him to handle a pre-trial evidentiary hearing. During this hearing, Williams, based on his review of the screening action form, announced to the court and Thompson’s counsel that the prosecution intended to test Thompson’s blood. However, no such test was ever ordered by the prosecution. Later, sometime before the April 11 trial,62 Dubelier asked Williams to take over trial responsibility for the case.

There is no testimony in the record from Deegan, who died several years before Thompson instituted this action. The record does contain, however, an affidavit and testimony from his colleague and close friend Michael Riehlmann. According to Riehlmann, shortly after Deegan was diagnosed with terminal cancer, Deegan confessed “that he had intentionally suppressed blood evidence in the armed robbery trial of John Thompson that in some way exculpated the defendant.”63 Deegan’s confession is, in part, supported by the evidence card, a card on file that identified the physical evidence in the case. The card indicates that the morning the trial was set to begin, Deegan checked out the blood evidence and never returned it.

As for the lab report itself, the record is not clear who ordered the testing. The report was dated April 9, just two days before trial, and addressed to Whittaker. Whittaker recalled seeing the report and placing it on Williams’s desk. Because he was only the screening prosecutor, Whit-taker was not responsible for turning it over to the defense. Williams and Dubelier both testified that they never saw the report. Riehlmann’s account of Deegan’s confession is not clear whether Deegan was referring to the lab report, to the actual blood evidence, or to both.

The statements of these four assistant district attorneys — the only prosecutors who had any involvement in the armed robbery case — provide very little information regarding the lab report. What is clear from the record is this: first, Whit-taker received the report a few days before trial; second, no prosecutor ever turned it over to Thompson’s counsel; and third, the report did not appear again until it was discovered fourteen years later.

*308Thompson based his case upon a single causation theory: that one or more of the assistant district attorneys involved in Thompson’s prosecution decided not to turn over the report because they did not know that they were legally obligated to produce it and that training sessions on Brady would have avoided this incident.64 To prove his theory, Thompson must present substantial evidence from which a jury reasonably could conclude that the failure of Connick to provide training sessions on Brady was the actual cause of and the moving force behind the failure to produce the report. The precedents require substantial evidence of direct causation. This standard demands more than evidence of confusion over Brady’s “gray areas” in the District Attorney’s Office. Finally, Thompson must establish that this lack of understanding would have been remedied by an in-house training program.

Thompson’s brief fails to point out any such evidence to sustain municipal liability. As best we understand his brief, the only arguments he makes regarding causation are these: (1) the record supports the conclusion that these four prosecutors knew about the blood evidence and yet failed to disclose it;65 and (2) the jury was free to reject Connick’s theory of a single rogue prosecutor. Even if we accept both of these assertions as correct, they still fail completely to establish that the Brady violation was caused by unfamiliarity with Brady. And because Thompson bore the burden of proof, he had to do more than simply assert that the jury was free to reject Connick’s explanations for the violation. Thompson had to put forth substantial evidence supporting his own theory of causation: that the assistant district attorney (or attorneys) responsible for the constitutional violation did not understand Brady, that this lack of understanding caused the failure to produce the report, and that Brady training could have resolved this lack of understanding.

We have reviewed the record for any such evidence. First, it contains evidence that Williams, when asked if Brady material includes documents that could be used to impeach a government witness, incorrectly replied “No.”66 Second, the 1987 policy manual from the District Attorney’s Office could be read to imply that Brady evidence need only be produced when the defense requests it and it fails to note that impeachment evidence is also included under Brady. Third, Solino and Connick, after the report was discovered, contended that the lab report was not subject to Brady as such because Thompson’s blood type was unknown and the report thus had no exonerative effect on Thompson’s guilt. Fourth, although Williams stated unequivocally that all technical or scientific reports, like the lab report, were required to be turned over to a defendant, he also testified that this obligation did not neces*309sarily arise from Brady because the report was not necessarily exculpatory.67

The record fails to establish, by substantial evidence, that the actual cause and moving force behind the constitutional violation of not producing the lab report was the failure of the District Attorney to have in-house training sessions on Brady. For example, an assistant district attorney’s confusion regarding whether Brady applied to impeachment evidence may show a need for enlightening this assistant but is irrelevant here because the lab report clearly was not impeachment evidence and would not have been turned over on that basis. The policy manual, although incomplete in its instructions on Brady evidence and post-dating Thompson’s trial by several years, does little to establish the necessary direct causal link, and the jury concluded in its verdict that the violation was not due to an established municipal policy.

Thus, even assuming that Connick was deliberately indifferent to a need for training, Thompson failed as a matter of law to show that the lack of training was the actual cause of the constitutional violation.68 Therefore the judgment should be reversed and rendered for the defendant.

C. Jury Instructions

The jury was probably misled in its decision by the district court’s plainly erroneous jury instructions. After several hours of deliberations, the jury sent out a single question:

What does “Deliberate” Indifference mean? Does it mean intentional or would “Failure to monitor” be considered Deliberate?

The district court responded that:

“Deliberate Indifference” does not necessarily mean intentional, but does require more than mere negilgence [sic] or even gross negligence.69

The district court’s answer lacked “concrete accuracy,”70 and instead defined de*310liberate indifference as something less than intent but more than negligence — a nebulous answer that failed to sufficiently inform the jury of the controlling law.71 Nor was there any clarification in the original jury instructions, which failed to state that municipal liability requires a “conscious” or “deliberate” choice on the part of the policymaker.

By failing to clearly define “deliberate indifference,” the instructions were confusing and misleading. The instructions erroneously omitted the clear and obvious requirement laid down by the Supreme Court in City of Canton that “[ojnly where a failure to train reflects a ‘deliberate’ or ‘conscious’ choice by a municipality — a ‘policy’ as defined by our prior cases — can a city be liable for such a failure under § 198S.”72 This en banc court has previously held that deliberate indifference is a “lesser form of intent.”73 The district court failed in its duty to incorporate these governing rulings when it issued jury instructions.

Although the plain error standard limits appellate review here because no proper objection was made to the instructions, we believe that standard was met.74 The district court’s failure to correctly instruct the jury was clearly and obviously inconsistent with the law, and this error affected the Defendant’s substantial rights by allowing liability without any actual municipal “culpability.”

Finally, plain error is to be corrected if the “error seriously affects the fairness, integrity or public reputation of judicial proceedings.”75 “The jury system is premised on the idea that rationality and careful regard for the court’s instructions will confine and exclude jurors’ raw emotions.”76 As noted above, the Supreme Court has specifically warned that reducing the standards of fault in municipal liability cases would “impose de facto respondeat superi- or liability” and raise “serious questions of federalism.”77 Correctly instructing the jury on the applicable standard of fault is particularly important in municipal liability cases which involve the public purses of our cities and local governments. In this case, nebulous jury instructions authorized a verdict manifestly unfair to these Defen*311dants and plainly inconsistent with the Supreme Court’s and our relevant precedent. The verdict also undermines the will of Congress expressed in § 1983.78

We urge this point in further explanation of this unjustifiable verdict and to discourage other district courts from making similar mistakes.

CONCLUSION

Judgment as a matter of law “is a method for protecting neutral principles of law from powerful forces outside the scope of law — compassion and prejudice.”79 We fully appreciate that Thompson has suffered a horrible wrong inflicted by agents of the government and that in many cases the principal would be responsible for the acts of these agents. But as Judge Wisdom counseled when overturning a jury verdict, “[i]n reviewing [a] ... case when the plaintiff has been injured grievously, hard as our sympathies may pull us, our duty to maintain the integrity of substantive law pulls harder.”80 The Supreme Court has stated clearly and emphatically that the liability of municipalities is limited to cases where a municipal action caused the constitutional violation. The plaintiff must show the “requisite degree of culpability” on the part of the municipality— deliberate indifference to an obvious need for training — and must demonstrate a “direct causal link” between the failure to train and the constitutional violation.81

Thompson failed to produce substantial evidence to support his claim that the District Attorney was deliberately indifferent to an obvious need for training of his staff. And he failed to produce adequate evidence of causation to show that the failure to train was the actual cause and moving force behind the failure to produce the lab report. Thompson has, in short, failed to meet the heightened standards for culpability and causation imposed by Monell, City of Canton, and Bryan County, and we would therefore reverse the district court’s judgment.

“The right of jury trial in civil eases at common law is a basic and fundamental feature of our system of federal jurisprudence which is protected by the Seventh Amendment. A right so fundamental and sacred to the citizen, whether guaranteed by the Constitution or provided by statute, should be jealously guarded by the courts.” Jacob v. City of New York, 315 U.S. 752, 752-53, 62 S.Ct. 854, 86 L.Ed. 1166 (1942).

The panel opinion thoroughly explains why the evidence the jury heard in this case is sufficient to support its verdict. See Thompson v. Connick, 553 F.3d 836 (5th Cir. 2008). Judge Clement’s dissent *312(“the dissent”) to this court’s order affirming based on a tie en banc vote, however, overlooks much of the evidence the jury heard and ignores the standard of review that we apply to jury verdicts.

By reading the dissent, one would be hard pressed to even realize that a jury rendered the verdict in this case. At the outset, the dissent attempts to explain the standard of review but fails to acknowledge the deference we must accord to a jury’s verdict. We have repeatedly admonished that

our standard of review with respect to a jury verdict is especially deferential. As such, judgment as a matter of law should not be granted unless the facts and inferences point so strongly and overwhelmingly in the movant’s favor that reasonable jurors could not reach a contrary conclusion.

Flowers v. S. Reg’l Physician Servs. Inc., 247 F.3d 229, 235 (5th Cir. 2001) (internal quotation marks and citations omitted). “A jury verdict must be upheld unless there is no legally sufficient evidentiary basis for a reasonable jury to find as the jury did.” Int’l Ins. Co. v. RSR Corp., 426 F.3d 281, 296-97 (5th Cir. 2005) (internal quotation marks omitted). We must view the evidence the jury heard with this deferential standard in mind.

A review of the full record — as laid out in the panel opinion — reveals that the dissent is merely quibbling with the jury’s factual findings. See Thompson, 553 F.3d at 843-46. This oversteps our bounds as an appellate court. The dissent presents nothing more than a skewed version of the facts in favor of the District Attorney’s Office. Its approach is directly contrary to the rule that we must view all evidence and draw all reasonable inferences in favor of the jury’s verdict. See United States v. Miles, 360 F.3d 472, 476-77 (5th Cir. 2004); Am. Cas. Co. of Reading, Pa. v. Myrick, 304 F.2d 179, 182 (5th Cir. 1962) (“[I]n reviewing a jury’s verdict a court may not substitute its judgment on the facts for the jury’s determination simply because inconsistent and uncertain inferences are equally supported by the proof.”). The dissent thus ignores the maxim that the jury is allowed to accept or reject competing evidence. See Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 150, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000) (noting that in reviewing a jury’s verdict an appellate court “may not make credibility determinations or weigh the evidence”).

For example, the dissent states that evidence regarding whether Connick had a policy to disclose crime lab reports was “uncontradicted and unimpeached.” See dissent at 306. However, ADA James Williams testified that he did not have a Brady duty to disclose crime scene technician reports and equivocated regarding the disclosure of blood reports if he did not know the perpetrator’s blood type, which is the exact situation he faced when prosecuting Thompson. R. at 2353-54 (suggesting that the crime lab report was not Brady material “because I didn’t know what the blood type of Mr. Thompson was, and I didn’t know what the blood type of Mr. LaGarde [the victim of the robbery] was”). Similarly, Williams explained that Connick had a policy not to disclose certain police reports and witness statements, contradicting his testimony that he had to turn over all written reports generated in a case. Compare R. at 2027, with R. at 2354. Val Solino, the DA’s Office’s official Rule 30(b)(6) representative, stated that under Connick’s Brady policy, even as later memorialized in a 1987 manual, an ADA in Connick’s office would not have had to produce a crime lab report if he did not know the defendant’s blood type. R. at 2874-75. As another example, the dissent minimizes the evidence the jury heard regarding the District Attorney’s other Brady violations in this and other cases. *313The dissent manages to highlight Con-nick’s testimony that the Louisiana Supreme Court had overturned only four convictions based on a Brady violation during his tenure as District Attorney, but fails to acknowledge Connick’s concession that because most Brady violations typically do not lead to published opinions, Brady violations had likely occurred in other cases as well. R. at 2823-25. Similarly, by discounting the numerous other examples of nondisclosure that occurred in this case (in addition to the nondisclosure of the crime lab report) — examples that Thompson presented to the jury without objection — the dissent ignores the fact that the jury could consider these other events1 as further proof of the need for and Con-nick’s indifference to training on Brady.

In another attempt to overturn the jury’s verdict, the dissent simply ignores evidence about causation. As discussed fully in the panel opinion, the evidence permitted the jury to find that Connick’s deliberate indifference caused the Brady violations in this case. See Thompson, 553 F.3d at 853-57. Thus, the only way the dissent can reach its desired result is by departing from our deferential standard, reading the record selectively to support its position, and substituting its own judgment for that of the jury’s.

Finally, the dissent acknowledges that Connick did not preserve his objection to the jury instructions but still attempts to support a conclusion that the district court plainly erred in its instruction on deliberate indifference. As the panel opinion explains, however, the district court’s instructions were legally correct— and certainly not so “clearly” or “obviously” wrong as to constitute plain error. See Thompson, 553 F.3d at 859-63.

Deliberate indifference and intent are not synonymous. They are instead separate, albeit sometimes legally equivalent, concepts. See Bd. of the County Comm’rs of Bryan County v. Brown, 520 U.S. 397, 419, 117 S.Ct. 1382, 137 L.Ed.2d 626 (1997) (“Deliberate indifference is thus treated, as it is elsewhere in the law, as tantamount to intent, so that inaction by a policymaker deliberately indifferent to a substantial risk of harm is equivalent to the intentional action that setting policy presupposes.”). One need not show actual intent for a jury to find liability under the deliberate indifference standard. In Hope v. Pelzer, 536 U.S. 730, 122 S.Ct. 2508, 153 L.Ed.2d 666 (2002), the Supreme Court clarified that the fact-finder can infer deliberate indifference “from the fact that the risk of harm is obvious.” Id. at 738, 122 S.Ct. 2508 (citing Farmer v. Brennan, 511 U.S. 825, 842, 114 S.Ct. 1970, 128 L.Ed.2d 811 (1994)); see also Farmer, 511 U.S. at 841, 114 S.Ct. 1970 (noting that the test from City of Canton v. Harris, 489 U.S. 378, 109 S.Ct. 1197, 103 L.Ed.2d 412 (1989), for deliberate indifference is an “objective standard”). This court also has stated that deliberate indifference is “a lesser form of intent.” Doe v. Taylor Indep. Sch. Dist., 15 F.3d 443, 453 n. 7 (5th Cir. 1994) (internal quotation marks omitted) (emphasis added). A finding of deliberate indifference is thus tantamount to a finding of intent in the context of municipal liability, but a finding of deliberate indifference does not require a finding of intent — which is exactly what the district court instructed *314the jury in light of this precedent. See Thompson, 553 F.3d at 859-63. The dissent improperly adds this intent element, thereby recasting the meaning of deliberate indifference so as to assign error to the district court.

At bottom, the dissent seeks to retry this case through the appellate process. This approach abdicates this court’s duty to uphold a jury’s verdict unless the facts point so strongly in Connick’s favor that no reasonable jury could rule to the contrary. See Flowers, 247 F.3d at 235. Indeed, the fact that reasonable judges on this court view the evidence differently suggests that these factual disputes were for the jury to resolve. As the extensive discussion in the panel opinion demonstrates, there was ample evidence to allow the citizens of this New Orleans jury to find for Thompson. Of course, this is an extraordinary case with extraordinary facts. Allowing this judgment to stand will not subject municipalities to widespread liability, as a holding that the need for training was “so obvious” and the lack of training “so likely” to create a constitutional violation will apply only in the rare instance. This is that rare case. The jury heard substantial evidence that the District Attorney’s Office provided no Brady-specific training, despite the known risk of the exact type of systemic nondisclosure that the failure to train caused here. Acknowledging the proper standard of review and viewing the jury’s verdict in the correct deferential light compels us to uphold the jury’s decision.