OPINION

delivered the opinion of the Court

Applicant was charged with the capital murder of his girlfriend’s 17-month-old child, Tristen Rivet. The State’s case largely depended on the expert opinion of Dr. Patricia Moore, the medical examiner who performed the autopsy and who testified that Tristen died from asphyxia due to the compression of her chest and abdomen. Applicant was convicted of capital murder and sentenced to imprisonment for life. His conviction was affirmed on appeal. Robbins v. State, 27 S.W.3d 245 (Tex.App.-Beaumont 2000). We affirmed the judgment of the court of appeals. Robbins v. State, 88 S.W.3d 256 (Tex.Crim.App. 2002). Since that time, Moore has re-evaluated her opinion and has stated that she can no longer stand by her trial testimony. Applicant filed this application for writ of habeas corpus, alleging actual innocence and due process claims. The convicting court entered findings of fact and conclusions of law. We will deny relief.

I. BACKGROUND

A. Trial

Applicant was indicted for the capital murder of 17-month-old Tristen Rivet. The indictment alleged that Applicant did “intentionally and knowingly cause the death of [Tristen] by asphyxiating [her].”1 The State elected not to seek the death penalty. The following facts were developed at trial.

The victim resided with her mother, Barbara Hope, and her mother’s boyfriend, Applicant, at the home of Applicant’s mother, Bonni Morris. Applicant and Hope had a volatile relationship, frequently separating and reuniting. Witnesses suggested that both suffered from depression. When seeking group-type counseling, Applicant told a counselor that he did not know what he would do if things got worse, and he feared he would hurt Hope if they stayed together.

Testimony indicated that Tristen and Applicant had a good relationship, but that changed in the months leading to Tristen’s death. Applicant’s personality began to change after he started taking pain medication for injuries received in a serious car accident. Then, beginning in November 1997, Tristen suffered injuries on three separate occasions while being cared for by Applicant: a bruise under the eye, an injury to her leg or ankle, and finally, a series of bruises across her face. Also, testimony suggested that in early 1998, Tristen became afraid of Applicant. Hope stated that Tristen “didn’t seem to care too much for [Applicant] anymore” and seemed afraid of him. Tristen’s injuries and change in behavior led neighbor Rhonda Bethune and babysitter Helen McDaniel 2 to express concern that Applicant was *449hurting Tristen. However, the defense presented several witnesses, including Morris and Applicant’s grandmother, brother, and sister-in-law, who stated that Tristen and Applicant had a very loving, father-daughter type of relationship.

On the morning of her death, Tristen was suffering from a cold but was otherwise in good health. Hope, accompanied by Morris, left the house at approximately 11:30 a.m. to attend appointments and run errands. Applicant was entrusted with Tristen’s care. Applicant’s parole officer, Tim Hurst, visited Applicant between 1:26 p.m. until 2:00 p.m. Hurst testified that he observed Tristen walking around and eating animal crackers, and Tristen asked for some red punch, which Applicant gave her from his own glass. Applicant’s brother arrived for a visit at approximately 1:45 p.m. and remained at the home until about 2:20 p.m.

Applicant paged Hope between 3:30 and 4:00 p.m. When Hope called, Applicant sounded “shaky” and “excited” and told her to hurry back to the house because he “had to go and had things to do.” When Hope and Morris arrived home between 4:00 and 4:30 p.m., Applicant told them that he had laid Tristen down for a nap shortly after they spoke on the telephone. Applicant stated that he had to leave, and an argument ensued with Hope about Applicant’s frequent absences. Applicant and Hope walked to the store a couple of blocks away and then returned home. During that time, Morris was alone with Tristen. She testified that she was going through bills and talking on the phone, as could be supported by phone records.

After Applicant departed, Hope watched a news broadcast on television. At about 5:40 p.m., Hope checked on Tristen and thought that the child was sleeping. At 6:00 p.m., Hope returned to Tristen’s room to wake her up. She saw that the baby was lying in her bed with a pillowcase covering one eye, part of her nose, and her mouth. When Hope moved the pillowcase, she saw that Tristen’s lips were blue. Upon picking her up, Hope found that Tristen’s body was cold and that she was not breathing.

Hope cried for Morris to call 9-1-1 for assistance and carried Tristen into the living room. There she held Tristen on her lap and tried to breath into her mouth. A pink fluid gurgled up from Tristen’s mouth and nose, and Hope inserted a finger into Tristen’s throat to attempt to dislodge any object stuck in her throat. Hope then carried Tristen outside, where she yelled for someone to assist her and placed the child on a patch of well-groomed lawn near the front door. Morris and a neighbor’s daughter, Pamela Garrison, attempted to perform CPR on Tristen. Morris blew into Tristen’s mouth while Garrison pushed with very little force upon the child’s abdomen three or four times, using the palm of her hand. Garrison testified that Tristen’s skin felt very cold, and she did not hear any air coming out of the baby. Another neighbor, Jackie Sullivan, who had previously worked as an emergency medical technician, approached and told Morris and Garrison to stop because they were performing CPR too forcefully, given the size of the child. Sullivan made a statement to the effect that they would kill the child if she was not dead already. She observed that Tristen was not breathing, that her body was cold, and that her lips were bluish-purple, circumstances leading her to believe that Tristen was dead at that time. Still, Sullivan started to perform infant CPR with two fingers.

*450An ambulance arrived at 6:08 p.m., and paramedic Elizabeth Fredregill placed Tristen on a stretcher. After several unsuccessful attempts, a breathing tube was inserted into Tristen’s larynx. Fire department personnel performed CPR and administered epinephrine during the trip to the hospital. Fredregill observed that Tristen was pale and cold to the touch, that her neck was stiff, and that there was vomit in her airway, and she formed an opinion that Tristen was dead based on her observation of fire department personnel performing CPR. The first base-line EKG was taken in the ambulance at 6:16 p.m.

Tristen arrived at the hospital at 6:36 p.m., and she was immediately examined by Dr. John Conner, who determined that Tristen was “asystole” and without respiration, was cool to the touch, and displayed some dependent lividity, all indicating that she “had been dead for some time.” Tris-ten was placed on monitors to assess her condition, but Conner believed that there was no chance of successful resuscitation. A nurse attempted to determine Tristen’s temperature with a rectal thermometer, which continued to display its lowest possible reading of 94 degrees Fahrenheit, thereby signifying that the child’s temperature was actually lower than the minimum displayed by the digital thermometer. Conner pronounced Tristen dead at 6:53 p.m. He broke the news to Hope, who was distraught and cried that she did not want to live. Conner testified that Applicant’s behavior was unusual in the situation because he attempted to fondle Hope, but other witnesses disputed this testimony.

Subsequently, Justice of the Peace Edie Connelly ordered an autopsy that was performed by the Harris County Medical Examiner’s Office (HCMEO), specifically assistant medical examiner Dr. Patricia Moore. Moore noted six or seven contusions on Tristen’s legs, which were consistent with normal childhood injuries. She also observed five irregularly shaped, purple bruises on Tristen’s back, ranging from one-eighth to one-quarter inch in width; bruises on the right side of her neck; and areas of discoloration on her face and left arm. Moore incised the bruises on Tris-ten’s back with a scalpel and found hemorrhages down to the level of deep subcutaneous tissue. When examining Tristen’s internal organs, Moore discovered petechi-ae (small areas of hemorrhage) on the thymus and the lungs, a small hemorrhage on the kidney, a recent hemorrhage between the intracostal muscles of the eleventh and twelfth ribs on each side, and a hemorrhage of the tonsils. Moore stated that Tristen’s heart appeared “pretty good” and the lungs contained “some mucus in the bronchi,” which probably resulted from a cold. Upon further examination the next day, Moore found two additional bruises behind Tristen’s right ear and another bruise on the right side of her neck.

At trial, Moore, as the State’s expert witness, testified that the cause of Tris-ten’s death was asphyxia due to compression of the chest and abdomen and that the manner of death was homicide. She explained that the presence of petechiae on the back of the thymus and lungs indicated an asphyxia-related death. Moore ruled out CPR as the cause of death because the injuries to Tristen’s back were inconsistent with the administration of adult CPR and the injury to the kidney was deep down, requiring a lot of force. She also excluded sudden infant death syndrome (SIDS) because of the child’s age “and the story doesn’t fit the picture of a SIDS baby death.” Additionally, Moore stated that Tristen may have been dead for at least three hours before her temperature was taken at the hospital, based upon an approximate post-mortem cooling rate of 1.5 *451degrees per hour, and that Tristen’s body would not have sustained bruises as the result of the application of CPR that long after her death.

On cross-examination, Moore testified that she was assuming that the CPR took place on the floor, so Applicant asked her to imagine that it took place on a floor with sticks and rocks scattered around and that the adults performing CPR were doing so as if Tristen was a strong adult and were applying heavy force on her chest and abdomen. Moore responded that such circumstances could explain the bruises on the back but not the rib injury (although she also acknowledged that if enough pressure was applied to the abdomen, the kidney and ribs could bruise). Moore also asserted that she was not completely excluding other reasonable hypothesis by which Tristen died. Still, it was her opinion that Tristen was asphyxiated, and she believed that beyond a reasonable doubt.

To contravene Moore’s testimony, the defense called Dr. Robert Bux, the deputy chief medical examiner for Bexar County, Texas.3 Bux testified that the cause of Tristen’s death could not be determined and that no anatomical reason demonstrated during the autopsy could have led to a specific cause of her death. Relying on a treatise, Bux explained that death from asphyxia by compression would have resulted in abundant petechiae above the level of compression (including on the forehead, cheeks, and eyes) as well as abrasions and bruises around the front of Tristen’s body that would have occurred during the struggle. But Moore observed none of these during the autopsy. Bux also stated that the occasional petechiae on internal organs observed were a “nonspecific finding” and could have resulted from other causes. That is, “[ejven the presence of abundant petechiae is not a hallmark” and was not “specific for asphyxia.” Regarding the time of death, Bux stated that pulseless electrical activity on Tristen’s EKG charts could have occurred 30 to 40 minutes after her death, indicating that Tristen’s death occurred after 5:30 p.m. when Applicant was not with the baby. When asked about the role of CPR in the injuries, Bux asserted that the bruises on Tristen’s back were consistent with the administration of CPR while the child was lying on a lawn that was not prepared for that purpose. Similarly, Bux testified that the rib and kidney injuries could have been caused by frontal pressure during CPR, although he admitted that he had not personally seen a kidney injury due to such occurrence. Bux claimed that bruises can occur after death, explaining that it is possible for there to be a distribution of blood vessels and then the blood runs out and pools because of gravity.

In its rebuttal case, the State offered evidence to contradict Bux’s EKG testimony. Fredregill described how the electrodes are attached and the advantages of electrodes over other types of monitors. Kelly Curry, Fredregill’s supervisor and the clinical manager for EMS at Montgomery County Hospital District who was trained in and also taught how to read EKGs, testified about interference and artifacts in EKG readings. She explained that the three electrode leads attached to Tristen looked at different directions of the heart and were to be read simultaneously. She dismissed as artifacts any activity on the EKG charts because no organized activity was represented and not all of the leads showed it. One reading did *452show some movement, but she attributed it to the CPR that was in progress, not heart activity. Finally, Carl Ulbrich, a physician who was working in the emergency room at the same time as Dr. Conner but was not the primary physician on Tristen’s case, reviewed the records and testified that the EKG readings indicated no electrical activity except for mild interference. He noted that the up-and-down pattern on one of the EKG charts was consistent with CPR.

Applicant testified in his defense. He stated that Tristen was affectionate toward him and that on the day of her death, he did nothing to harm Tristen. In fact, he claimed that he had never struck her, abused her, disciplined her, or even raised his voice to her. Yet he admitted causing the three injuries to Tristen, blaming the incidents on his “carelessness.” When asked about his turbulent relationship with Hope, Applicant denied that any stress resulted from the fact that Hope came in and out of his life, and although he participated in group counseling with her, he denied being depressed, claiming he attended merely out of concern for Hope. Applicant further commented that he would overlook a lot of things because of his love for Tristen. In addition, Applicant was questioned about the testimony of Bethune that in the month following Tristen’s death, Applicant took all of the batteries out of Tristen’s toys and then explained to her that “it hurt too much; he couldn’t handle the guilt.” Applicant responded that he removed the batteries because it hurt him to hear them go off, not due to a feeling of guilt from something that he had done.

During closing arguments, the State emphasized Moore’s testimony in arguing that it was Applicant, and only Applicant, who could have caused the asphyxia-related death of Tristen. For his part, Applicant argued that if “anything, he is guilty of the offense of loving a child.” Applicant put forth the SIDS scenario and emphasized that the bruises on Tristen’s body could have been caused by incorrectly performed CPR efforts to save her life. He also pointed to the testimony of the two medical examiners, arguing that Bux’s was the more credible opinion.

On February 22, 1999, the jury found Applicant guilty of capital murder, and Applicant was sentenced to life imprisonment. Approximately a month later, Applicant filed a motion for new trial, arguing that evidence was legally and factually insufficient to establish that Tristen’s death was a homicide, but the trial court denied the motion.

B. Beaumont Court of Appeals

Applicant argued on appeal, inter alia, that the evidence was legally and factually insufficient to establish that he caused Tristen’s death. The court of appeals rejected Applicant’s arguments and affirmed the conviction. Robbins v. State, 27 S.W.3d 245 (Tex.App.-Beaumont 2000). In overruling Applicant’s legal-sufficiency claim, the court of appeals explained that the State presented evidence such that a “rational trier of fact could find [Applicant] committed capital murder as alleged in the indictment.” Id. at 247-48. For the factual-sufficiency claim, the court of appeals noted that Applicant developed alternative theories that either Hope intentionally killed Tristen after Applicant left the home or the women accidentally killed the child while trying to revive her from an unexplained coma, and the court concluded that the “evidence supporting these alternate hypotheses is not so overwhelming that [it] could conclude that the verdict is contrary *453to evidence.” Id. at 249.4

C. Court of Criminal Appeals

We granted a petition for discretionary review to determine whether the trial court properly admitted evidence of previous injuries that Tristen suffered while she was in Applicant’s care.5 We affirmed the court of appeals. Robbins v. State, 88 S.W.3d 256 (Tex.Crim.App. 2002).

We first rejected Applicant’s Rule 404(b) claim that the relationship evidence was improperly admitted to show intent or absence of accident because his plea of not guilty did not put his intent at issue. Id. at 259-62. While his plea alone did not put his intent at issue, Applicant “went beyond simply pleading not guilty through vigorous cross-examination of the prosecution witnesses suggesting that the victim’s death was caused by some means other than [hisj intentional act.” Id. at 261. Therefore, we determined it was “at least debatable” whether Applicant’s intent was at issue, and the trial court would not have erred in finding that the relationship evidence was relevant to Applicant’s intent or to “the noncharacter conformity purpose of rebutting [Applicant’s] various defensive theories.” Id. at 261-262.

We then overruled Applicant’s Rule 403 argument that the relationship evidence’s probative value was substantially outweighed by the danger of unfair prejudice. Id. at 262-63. We determined that the relationship evidence, though prejudicial, was not “unfairly prejudicial.” Id. at 263. “[T]here was no reason to believe that the jury had a reasonable doubt for [Applicant’s] guilt of the charged offense but convicted [him] based on the relationship evidence.” Id.

D. Reevaluation of Autopsy Findings

In March 2007, an acquaintance of Applicant contacted the HCMEO and asked it to review Moore’s findings regarding the cause of Tristen’s death. The deputy chief medical examiner for Harris County, Dr. Dwayne Wolf, undertook a re-evaluation of the autopsy findings. After reviewing the testimony adduced during Applicant’s trial, the autopsy report, the EMS and medical records, and the police offense report, Dr. Wolf concluded that Moore’s observations during the autopsy did not support a finding that the death resulted from asphyxiation by compression or from any other specific cause. Consequently, on May 2, 2007, Wolf amended Tristen’s autopsy report to reflect that both the cause and manner of death was “undetermined.” And so on the following day, Justice of the Peace Edie Connelly formally reopened the inquest into Tristen’s death.

Shortly thereafter, former Harris County Medical Examiner Joye Carter was asked by the Montgomery County District Attorney’s Office to review Moore’s autopsy report. Carter had been Moore’s supervisor and had agreed with Moore’s *454original opinion. In a May 10 letter to the district attorney, she wrote, “Upon my review of this case I would not concur with the opinion on the manner of death as a homicide but would reconsider this case as an undetermined manner,” and “If the Harris County Medical Examiner intends to re-rule this case as an undetermined manner of death I would agree with that change.”

Moore, too, was asked by the Montgomery County District Attorney’s Office to review her autopsy report. In a May 13 letter to the district attorney, she stated,

I believe that there are unanswered questions as to why the child died, and I still feel that this is a suspicious death of a young child. Given my review of all the material from the case file and having had more experience in the field of forensic pathology, I now feel that an opinion for a cause and manner of death of undetermined, undetermined is best for this case.

Moore explained that since her original opinion, she has had more experience, and she has reviewed additional information that suggested that the bruises could have resulted from aggressive CPR and other efforts to assist the child.6 She emphasized that it was significant that aggressive adult-type CPR by untrained persons was performed on Tristen, a 17-month-old child.

E. Habeas Application and Proceedings

On June 4, 2007, Applicant filed his original application for a writ of habeas corpus alleging, “Newly discovered evidence shows that no rational juror would find Applicant guilty beyond a reasonable doubt of the offense for which he was charged and convicted.” About a month later, Applicant filed a supplemental application alleging that his “right to a fair trial by a fair and impartial jury ... was violated because his conviction was based on testimony material to the State’s case that has now been determined to be false.”

In its original response on June 25, 2007, the State recommended that Applicant be granted a new trial because his due process rights to a fair trial and impartial jury were violated. The State claimed that because it relied on Moore’s original opinion in presenting its case, which has now been recanted, confidence in the outcome has been undermined. Citing Ex parte Carmona, 185 S.W.3d 492 (Tex.Crim.App. 2006), the State wrote, “While Dr. Moore’s testimony is not perjured testimony, the effect of the change in her opinion is the same — the jury was led to believe and credit facts that were not true.”

The same day, Applicant and the State filed agreed findings of fact and conclusions of law. But instead of signing them, on August 22, 2007, the trial court appointed Dr. Thomas Wheeler, the Chairman of the Department of Pathology at Baylor College of Medicine, with the task of conducting an independent pathological examination to address the following issues: (1) What is the manner of Tristen Rivet’s death? (2) What is the means of Tristen Rivet’s death? (3) Are the manner and *455means of Tristen Rivet’s death able to be determined? (4) Does a change in the medical examiner’s opinion about the manner and means of Tristen Rivet’s death entitle Applicant to a new trial? After reviewing the autopsy file of the victim, trial testimony, and exhibits, Wheeler concluded in a September 18, 2007 letter to the trial court that the cause and manner of Tristen’s death were undetermined. Wheeler asserted that “[a]lthough the autopsy performed by Dr. Moore was thorough and well documented, her conclusion that the death of Tristen Rivet was caused by asphyxia secondary to chest compressions was not justified by the objective facts and pathological findings in this case.” He could not rule out suffocation or asphyxiation as the cause of death, but he did not see any physical findings that would support any particular conclusion as to the cause of death.

In October 2007, Judge Connelly ordered that pathologist Linda Norton conduct an independent forensic examination of the evidence and submit a written report of her findings and opinion on the cause and the manner of Tristen’s death.7 On March 28, 2008, Norton reported the results of her review during a recorded telephone conference call, in which Judge Connelly and counsel for Applicant and the State participated. Norton stated that it was her opinion that Tristen’s death was a homicide and that the manner of death was asphyxia by suffocation. She explained that her conclusion was supported by the petechial hemorrhages on Tristen’s lungs and thymus, combined with the other evidence of trauma, and in the context of the other circumstances of Tristen’s death. In addition, Norton stated that the correct rule of thumb for assessing temperature loss in a child’s body after death is an approximate loss of three degrees per hour, depending upon ambient temperature and other environmental facts. Thus, combining that with Tristen’s maximum rectal temperature of 94 degrees at the hospital and the descriptions of Tristen’s condition by Sullivan and others, she believed that Tristen’s death occurred between 2:30 and 5:00 p.m. Consequently, because the child had been dead for at least an hour before CPR was attempted, the external bruises observed during the autopsy could not have been inflicted during the CPR. Nonetheless, Norton acknowledged that she could not conclude beyond a reasonable doubt that Applicant and Applicant alone committed the homicide.

Norton also recommended that authorities investigate reports that Applicant had written something on a dollar bill and placed it in Tristen’s casket at the funeral home on the date of Tristen’s funeral. Ruth Hope (Barbara Hope’s mother) and Shelby Becker (Barbara Hope’s sister) had executed affidavits indicating that they saw Applicant writing something on a money bill and then placing it in Tristen’s coffin. On April 4, 2008, Judge Connelly signed an order directing that Tristen be exhumed for the purpose of retrieving any evidence that might be found in the casket. Six days later, Tristen’s remains were exhumed and remnants of a piece of paper resembling United States currency were recovered from the casket liner. Document preservation experts reported on May 6, 2008, that no markings of any kind could be identified due to the poor condition of the paper.

Judge Connelly amended Tristen’s death certificate on May 13, 2008, to correspond *456with Norton’s opinion that Tristen’s death was caused by asphyxia due to suffocation, rather than asphyxia by compression; the homicide finding was not changed. The following day, Norton executed an affidavit incorporating her conference discussion. In its May 27, 2008 supplemental response, the State was no longer willing to recommend a grant, but it agreed not to oppose Applicant’s request for a new trial. It wrote that the “cause of death remains asphyxiation, albeit by suffocation rather than compression, and the manner of death a homicide as presented by the jury at Applicant’s trial.” On August 6, Wheeler submitted a sworn affidavit, repeating what he had said in his September letter to the trial court, adding that he disagreed with Norton’s opinions. That same day, Applicant filed a memorandum on why Moore’s amended autopsy should not be found credible.

On August 13, 2008, Applicant, joined by the State, filed proposed joint findings of fact and conclusions of law, which recommended that Applicant be granted a new trial based on due process grounds and the fact that he was denied a “fundamentally fair trial and an accurate result.” On August 25, Moore’s sworn affidavit was filed, incorporating much of her prior letter to the district attorney. The next day, Applicant, again joined by the State, filed another set of proposed findings and conclusions. Instead of signing it, the trial court ordered that the parties engage in discovery. Moore was deposed on December 10, 2008; Wheeler on December 19, 2008; and Wolf on February 10, 2009. The trial court appointed John Milutin, an attorney experienced in the deposition of medical experts, to depose these witnesses. Norton was subpoenaed so she, too, could be deposed, but she could not be located. When the trial court granted the State’s motion to depose Norton at the location of her choosing, she could not be deposed due to medical problems.8 Instead, on December 17, 2009, Norton submitted a second affidavit in which she confirmed that she was incapable of preparing for or participating in a deposition, and she adopted and ratified under oath the statements and opinions she expressed during the previous telephone conference, including that she believed Tristen died from suffocation and that her death was homicide. Based largely on Norton’s opinion, on December 22, the State filed its second supplemental response and recommended that relief be denied. Shortly thereafter, Applicant filed an objection to Norton’s affidavit, arguing that, given her unwillingness to be deposed, the trial court should not consider her affidavit.

On December 29, 2009, Judge Connelly conducted an evidentiary hearing on Applicant’s motion to reopen the inquest into Tristen’s death to allow consideration of additional expert medical testimony. Applicant also filed a motion to reopen the inquest into Tristen’s death after her death certificate was amended to show she died of suffocation, but this motion was denied because “on the basis of examination and investigation, in the opinion of *457this Court, the cause and manner of the death of Tristin Skye Rivet, as shown on the amended death certifícate ..., is cause: asphyxia due to suffocation, manner: homicide.”

On January 15, 2010, the State filed its proposed findings of fact and conclusions of law, which recommended that relief be denied. Days later, Applicant filed his proposed findings and conclusions. On January 21, the State filed its first supplemental brief in support of its proposed findings and conclusions. While not willing to concede that Applicant properly raised a due process claim in his supplemental ground for relief, the State argued that, even if he did raise due process, the Court “has not yet held — and it seems unlikely that it will ever hold — that the Due Process Clause is violated when a witness provides, in good faith, an opinion that is believed to be true by both the witness and the prosecution at the time of trial, even if that opinion is subsequently challenged by other experts or reconsidered by the witness who offered it.”

The next day, on January 22, 2010, the trial court permitted oral argument. Applicant argued that Moore’s re-evaluation was newly available evidence and that Ex parte Elizondo, 947 S.W.2d 202 (Tex.Crim.App. 1996), requires that the newly available evidence be evaluated within the four corners of the trial transcript.9 Further, Applicant asserted that due process and fairness require that the jury have the opportunity to re-weigh the evidence. In contrast, the State contended that Applicant could not establish that he was actually innocent because the evidence is not newly discovered, the re-evaluation was not indisputable, and there was other evidence of Applicant’s guilt. Regarding the due process claim, the State argued that Applicant had failed to raise it as a supplemental ground, and it doubted whether there was a legal and factual basis for his due process claim: “It’s hard to believe that a violation of due process is established by evidence that an expert opinion may have been correct or it may have been incorrect.”

The trial court made twenty-two pages of detailed findings of fact, much of which is summarized above, and five pages of conclusions of law. The trial court recommended that we grant Applicant a new trial because his due process and due course of law rights were violated, as was his right to an impartial jury. But it recommended that we deny Applicant’s actual innocence claim. In a post-conviction review of a writ of habeas corpus, this Court is the ultimate factfinder. We are not bound by the findings and conclusions of the convicting court, but we generally defer to such if they are supported by the record. See, e.g., Ex parte Chabot, 300 S.W.3d 768, 772 (Tex.Crim.App. 2009); Ex parte Thompson, 153 S.W.3d 416, 417-18 (Tex.Crim.App. 2005).

II. ACTUAL INNOCENCE

Applicant contends in his original ground for relief that “[njewly discovered evidence shows that no rational juror would find Applicant guilty beyond a reasonable doubt of the offense for which he was charged and convicted.”

A. Caselaw

Although a strong presumption of finality to a criminal conviction exists, it must sometimes yield so that a “fundamen*458tally unjust incarceration” may be corrected. Ex parte Franklin, 72 S.W.3d 671, 678 (Tex.Crim.App. 2002). Still, any person finally convicted in a fair trial cannot wage a collateral attack “without making an exceedingly persuasive case that he is actually innocent.” Ex parte Elizondo, 947 S.W.2d 202, 206 (Tex.Crim.App. 1996).

We have recognized that a claim of actual innocence is cognizable in a post-conviction habeas corpus proceeding. Elizondo, 947 S.W.2d at 205. There are two types of actual innocence claims that may be raised. One is a Herrera-type claim or a bare claim of actual innocence. Id. at 208. It is a substantive claim in which the applicant argues his innocence based solely upon newly discovered evidence, evidence that was neither introduced at trial nor available to the defense to introduce at trial. Id. Because this attacks the conviction directly, an “extraordinarily high” standard applies. Id. at 209; see Schlup v. Delo, 513 U.S. 298, 315-16, 115 S.Ct. 851, 130 L.Ed.2d 808 (1995). The second type is a Schlup-type claim, which is accompanied by a claim of constitutional error. Schlup, 513 U.S. at 314, 115 S.Ct. 851. The applicant’s claim of innocence does not provide a basis for relief by itself.

B. Application

Applicant presents a bare innocence claim, relying on the newly available evidence of Moore’s re-evaluation. In assessing Applicant’s claim, we look to see whether there is “clear and convincing evidence that no reasonable juror would have convicted him in light of the new evidence.” Elizondo, 947 S.W.2d at 209. We evaluate “the probable impact of the newly available evidence upon the persuasiveness of the State’s case as a whole, [so] we must necessarily weigh such exculpatory evidence against the evidence of guilt adduced at trial.” Id. at 206; see Ex parte Thompson, 153 S.W.3d at 417.

The convicting court concluded that Applicant has failed to meet the burden of showing his actual innocence and recommended that this Court deny relief based on this ground. Because we find support in the record, we agree with the convicting court.

Even assuming that Moore’s re-evaluation was - previously unavailable, Applicant has failed to prove that the new evidence unquestionably establishes his innocence. Moore can no longer stand by her trial testimony, but rather than completely retracting her trial opinion, she is of the current opinion that the cause and manner of Tristen’s death are “undetermined.” Moore cannot rule out her trial opinion as a possibility of how Tristen died. Hence, Moore’s re-evaluation falls short of the requisite showing for actual innocence because it does not affirmatively disprove that Applicant intentionally asphyxiated Tristen.

Our conclusion is evident when Moore’s opinion of “undetermined” is placed in the context of the trial record, including Bux’s “undetermined” opinion (which the jury heard at trial and did not believe) and the non-medical evidence. For example, Applicant and Hope had a stormy relationship, which led both to seek counseling. Applicant evidenced a personality change that accompanied his use of pain medication in the months preceding Tristen’s death. Tristen became afraid of Applicant, cowering in his presence, and suffered at least three injuries when in Applicant’s sole care.10 Applicant was alone with Tristen for much of the day, and she *459was in good health when visitors stopped by the house. Applicant seemed to be in a rush to leave the house when Hope returned home. Despite Tristen’s general good health, she died suddenly. Applicant spoke of feelings of guilt while subsequently discussing Tristen’s death, and her toys, with a neighbor. In all, when the newly available evidence is viewed in the context of the inculpatory evidence in the trial record, it cannot be said that Applicant’s innocence has been unquestionably established because a juror could reasonably conclude that Applicant intentionally asphyxiated Tristen.

The case at hand is distinguishable from our leading actual innocence case of Elizondo, 947 S.W.2d 202. There, the applicant’s conviction was based solely on the victim’s testimony; no other incriminating evidence was offered at trial. Id. at 209-210. The victim later recanted his trial testimony, claiming that it was entirely false, and we concluded that there was clear and convincing evidence “that another jury hearing the [newly available] evidence ... would view the new evidence as the more credible and would acquit applicant.” Id. at 210. The recantation “not only void[ed] his trial testimony which implicates applicant, but constitute^] affirmative evidence of applicant’s innocence.” Id.

Unlike the witness in Elizondo, Moore did not entirely repudiate her testimony. Although she can no longer stand by her more definite trial testimony, it remains at least possible that Tristen’s death could have occurred as Moore originally testified. Thus, Moore’s re-evaluation does not void her trial testimony. The jury could have considered Moore’s “undetermined” opinion and still found Applicant guilty, especially in light of all of the other evidence adduced at trial. Applicant has, therefore, failed to make the requisite threshold showing “by clear and convincing evidence that no reasonable juror would have convicted him in light of’ Moore’s re-evaluation. See id. at 209.

III. FALSE TESTIMONY AND DUE PROCESS

Applicant alleges in his supplemental ground for relief that his “right to a fair trial by a fair and impartial jury as guaranteed by the Sixth and Fourteenth Amendments to the United States Constitution, and Art. I, Sec. 10, of the Texas Constitution, was violated because his conviction was based on testimony material to the State’s ease that has now been determined to be false.”

A. Caselaw

The Due Process Clause of the Fourteenth Amendment can be violated when the State uses false testimony to obtain a conviction, regardless of whether it does so knowingly11 or unknowingly12. Accordingly, to constitute a due process violation, the testimony used by the State must have been false, and it must have been material to the defendant’s conviction, meaning “there is a reasonable likelihood 13 that the false testimony could have affected the judgment of the jury.” United States v. Agurs, 427 U.S. 97, 103-04, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976); Ex *460parte Ghahremani, 332 S.W.3d 470, 478 (Tex.Crim.App. 2011).

The case law in this area frequently refers to the use of perjured testimony, but it is sufficient that the testimony was “false.” Napue, 360 U.S. at 269, 79 S.Ct. 1173. There is no need to show that the witness committed perjury, although whether the testimony is perjury or was knowingly used may impact the applicable standard of materiality. Ghahremani, 332 S.W.3d at 478. We have explained that “‘[testimony that is untrue’ is one of many ways jurists define false testimony [and the] Supreme Court has indicated that ‘improper suggestions, insinuations and, especially, assertions of personal knowledge’ constitute false testimony.” Ex parte Chavez, No. AP-76291, 2010 WL 4638619, at *7 (Tex.Crim.App. Nov. 17, 2010) (not designated for publication) (quoting Berger v. United States, 295 U.S. 78, 88, 55 S.Ct. 629, 79 L.Ed. 1314 (1935) and Black’s Law Dictionary 1485-86 (7th ed. 1999)).

B. Application

Applicant alleges that Moore’s trial testimony was false. The convicting court, in its conclusions, determined that “[t]he verdict was not obtained by fair and competent evidence, but by admittedly false testimony that was unsupported by objective facts and pathological findings and not based on sufficient expertise or scientific validity.” We disagree with the convicting court because the record does not support its conclusion.

At trial, Moore testified that it was her professional opinion that the cause of death was asphyxia-related compression of the chest. In explanation, Moore relied heavily on the petechiae found on the thymus and lungs, which indicated some type of asphyxia, and the small hemorrhage on the kidney, which suggested some type of injury. On cross-examination, she acknowledged that she was not completely excluding other reasonable hypothesis by which Tristen died. Rather, it was her opinion beyond a reasonable doubt that Tristen’s death was due to homicide by asphyxiation.

Moore has since changed her position and stated that she can no longer stand by the position to which she testified.14 In her sworn affidavit, she asserted, “I now feel that an opinion for a cause and manner of death of undetermined, undetermined is best for this case;” yet, she felt that “this is a suspicious death of a young child.” During her deposition, Moore affirmed that she “could no longer testify within a reasonable degree of medical certainty that the complainant’s death in this case was the result of compression asphyxia” or “that the [manner] of death in this case was homicide.” Still, she could not exclude such cause and manner of death, particularly because “the diagnosis of as-phyxial death ... relies on investigations since anatomic findings are typically subtle or absent.” She maintained that the cause and manner of Tristen’s death should now be considered undetermined:

Q. ... So nothing about that would tell you that this baby’s manner of death was homicide from what you — what you found on autopsy?

A. Well, back then I did believe that the bruising on the back and the chest compressions was what caused the baby; so, I believed that was caused by another person. That’s what I believed back then, but I don’t believe that now.

Q. What do you believe caused it now?

A. I’m not sure.

*461Q. What do you believe caused it? I’m not asking what you’re sure of.

A. I don’t know.

We believe that the record does not support that Moore’s trial testimony has been proven to be false. Although Moore played an important role in the State’s case as the only trial witness to point to a specific cause of Tristen’s death, Moore’s trial testimony is not false just because her re-evaluation of the evidence has resulted in a different, “undetermined” opinion, especially when neither she nor any other medical expert can exclude her original opinion as the possible cause and manner of death.

This case is similar to our recent unpublished opinion of Ex parte Chavez. There, a criminalist testified at trial that an unknown hair from tape lifted at the victim’s residence shared similar physical characteristics with the defendant’s known hair sample, but she could not positively identify the unknown sample. Chavez, 2010 WL 4638619, at *7. The similarities of the physical characteristics of the unknown hair and defendant’s known hair sample were not refuted, but subsequent DNA testing excluded defendant as a source of the unknown hair. Id. We held that, even with the new DNA results, the criminalist’s testimony was not false. Id.

Here, similar to the criminalist’s testimony that she could not positively identify the sample, cross-examination by the Applicant established that Moore’s testimony was her professional opinion and that she was not ruling out other reasonable hypothesis by which Tristen died. In addition, like the criminalist’s testimony, that asphyxia was the cause and homicide the manner of Tristen’s death has not been entirely refuted. As the convicting court determined, “[n]o expert rules out asphyxia as the cause of death,” “[n]o expert can exclude Applicant as the perpetrator if it is a homicide, and no expert has excluded homicide as the manner of death.” During the habeas proceedings, various experts have opined that the autopsy findings do not adequately support Moore’s conclusion that the death was a homicide by asphyxiation (and Moore herself has adopted that position15), but none of the experts have stated that Tristen could not have been intentionally asphyxiated. And although they critique Moore’s interpretation of the petechiae evidence upon which she relied at trial, the “non-specific” indicator cannot be ruled out as being the result of asphyxiation. On the other hand, at least one well-qualified pathologist, Dr. Norton, has concluded that the child was a victim of homicide by asphyxiation.16

*462We must note that the convicting court concluded that “[a]t the time [Moore’s] opinions were sworn to, they were false or at least gave a false impression.” It is true that we have held that testimony may be false because it creates a false impression of the facts. See Ghahremani, 332 S.W.3d at 477-78 (“It is sufficient if the witness’s testimony gives the trier of fact a false impression.”). However, Moore’s reevaluation of her testimony can be distinguished from these instances of false testimony.

Generally, when we have held that testimony is false because it creates a false impression, the witness omitted or glossed over pertinent facts. For example, in Alcorta v. Texas, 355 U.S. 28, 78 S.Ct. 103, 2 L.Ed.2d 9 (1957), the defendant claimed that he murdered his wife as a result of sudden passion when he came home to find a man kissing her. The man, the only eyewitness to the murder, testified that his relationship with the wife was nothing more than a casual friendship in which he drove her home from work a few times. The Supreme Court held that such testimony was false; it created a false impression of the facts because, in actuality, the man was the wife’s “lover and paramour,” and the two had sexual intercourse on many occasions. Id. at 30-32, 78 S.Ct. 103. Similarly, in our recent case of Ghahremani, 332 S.W.3d 470, the victim’s mother testified that after the victim was assaulted, the victim “gained weight, required therapy, and “wasn’t the same socially,’ ” but the mother denied that anything else was different about the victim during the time between the assault and the victim’s intensive therapy.17 Id. at 478-79. However, the habeas record showed that the victim sold drugs and was initiated into a gang during that period. We concluded that the witness’s testimony created a misleading impression of the facts because of the gravity of the events omitted, the significant period of time that was not addressed, and the fact that the testimony attributed all of the psychological treatment to the defendant’s actions. Id. at 479; see also Burkhalter v. State, 493 S.W.2d 214, 218 (Tex.Crim.App. 1973) (holding that, although the witness’s statement was not technically false, it “conveyed an impression to the jury which the State knew to be false” when the witness’s lawyer had an understanding with the State that the witness would not be prosecuted if he testified, but the witness testified he did not have an agreement with the State).

In contrast to those cases, Moore’s trial testimony did not result in a false impression of the facts. Unlike the witness in Alcorta, Moore testified openly about the autopsy findings and her professional opinion regarding the cause and manner of Tristen’s death, and unlike the witness in Ghahremani, Moore did not omit pertinent details. As discussed previously, at trial, Moore explained the reasoning behind her original conclusion that Tristen’s death was asphyxia-related. Also, neither Moore’s conclusion nor the autopsy evidence upon which she relied has been entirely refuted by any expert. Accordingly, we do not believe that Moore’s testimony was false; it did not create a false impression of the facts.

Applicant relies upon Ex parte Henderson, 246 S.W.3d 690 (Tex.Crim.App. 2007) (per curiam), in advancing his *463due process claim, but his argument is misplaced. There, the medical examiner testified at trial that it would have been “impossible” for the victim’s injuries to have occurred as the defendant claimed. Id. at 691. Advances in scientific research subsequently indicated that it was possible for the injuries to have occurred in the claimed manner, so the medical examiner filed an affidavit stating that he could no longer stand by his trial testimony. We held that his re-evaluation was a “material exculpatory fact.” Id. at 692. Significant to this case is that, although we granted a request for the stay of defendant’s execution and remanded the cause for an evi-dentiary hearing pursuant to a subsequent writ application, we did not find that any due process violation had occurred or that the defendant had succeeded in demonstrating her actual innocence. Thus, Applicant here may not rely on that case for such a proposition (i.e., that the admission of expert medical opinion testimony constituted a due process violation because the expert subsequently repudiated the opinion he offered at trial). Also, unlike Henderson, Moore’s re-evaluation was not attributed to new scientific information but was instead based on the same autopsy findings and other information that she had at the time of trial.

For these reasons, Moore’s trial testimony has not been proven false. Thus, the State did not use false evidence to obtain Applicant’s conviction, and Applicant does not have a due process right to have a jury hear Moore’s re-evaluation.

IV. CONCLUSION

We conclude that Applicant has failed to establish that newly available evidence has unquestionably established his innocence under Elizondo and that it has not been proven that the State used false testimony to obtain Applicant’s conviction. Relief is denied.

PRICE, J., filed a concurring opinion in which MEYERS and HERVEY, JJ., joined.

The applicant raises two claims in this post-conviction application for writ of ha-beas corpus: actual innocence and false evidence. Each type of claim is cognizable in the post-conviction habeas context and, should they pan out, each could serve independently to defeat the State’s otherwise legitimate interest in the finality of its convictions. But, though each finds its basis in the Due Process Clause of the Fourteenth Amendment, the two types of claims suffice to defeat the State’s finality interest for different reasons.

A claim of actual innocence is cognizable, we have recognized, because due process will not tolerate stigmatizing and punishing an individual who can definitively demonstrate that he did not commit the crime for which he has been convicted.1 Thus, actual innocence is concerned with the accuracy of the original proceeding. If the applicant can produce new evidence that affirmatively demonstrates his innocence (and does not simply call the State’s original evidence of his guilt into question), and, factoring that new evidence in with the trial evidence, the habeas court concludes by clear and convincing evidence that no rational jury would have convicted *464him, he is entitled to relief.2 That is a “Herculean” burden, we have said,3 and it is meant to be. Otherwise, trials would cease to be the so-called “main event” of our adversarial criminal justice system,4 and the State would never be able to depend upon the finality of its convictions.

A claim of false evidence, on the other hand, is predicated less on due process concerns with the bottom-line accuracy of the trial result and more on fundamental due process guarantees of fairness in the main event itself.5 If we are to afford the result of a criminal trial a presumption of finality, then we must be able to assure defendants that, as the main event, the trial will be conducted in as fair a manner (less than absolutely perfect) as we can muster. There is nothing fair about a trial in which it can be shown that the State deliberately utilized perjured, fabricated, false, or even just plain misleading evidence to obtain a conviction. Even a defendant who is absolutely guilty — assuming we could ever objectively know such a thing, apart from the convention of an institutional factfinder’s verdict — may obtain a new trial if his original trial was tainted by such glaring unfairness and (in the habeas corpus context) he can establish that the unfairness more likely than not contributed to his conviction.6

But what if neither the State nor the defendant was aware that some material evidence that was admitted at trial was perjured, fabricated, false, or misleading? In the last several years, we have held that a habeas applicant may still obtain relief under these circumstances, utilizing the same standard of harm (for obtaining relief in state habeas corpus proceedings) as applies to the State’s knowing use of such evidence.7 It is not entirely clear to me, however, in the context of the inadvertent use of such evidence, whether the primary due process impetus is to assure the fairness of the process or the accuracy of the result (or maybe a little of each). Is it that it strikes us as unfair to uphold a conviction that is based upon false evidence regardless of the source of the evidence or the motive behind its admission? In other words, is misconduct or malfeasance on the State’s part unnecessary to our perception that the process has not operated fairly? Or are we more concerned that a conviction based upon what later turns out to be false evidence does not provide sufficient assurance that we are stigmatizing and punishing a truly guilty defendant? While these concerns are obviously interrelated, they are, nonetheless, distinct. And it is important, in defining whether a particular due process claim is compelling enough to defeat the State’s presumptive interest in finality, to understand the distinction. Knowing how *465to set the appropriate boundaries for what we are willing to call “false” evidence in the first instance largely depends upon which due process rationale — fairness or accuracy — is predominant.

I think it is fair to say that the State’s inadvertent use of false evidence is not as unfair as the State’s deliberate use of false evidence. So the justification for holding that the State’s inadvertent use of false evidence may defeat its otherwise legitimate interest in finality must hinge more on a concern for the accuracy of the result. And yet, before a concern for accuracy may, by itself, justify undoing a criminal conviction, we have said that the habeas applicant must satisfy the “Herculean” burden of establishing his actual innocence.8 Unless we are to allow our false evidence jurisprudence unduly to encroach upon our actual innocence jurisprudence, it seems to me that when it comes to claims of the inadvertent use of false evidence, we must not be overly liberal in how we characterize “false” evidence.

And that brings me to the facts of the instant case.9 Dr. Moore has renounced her trial testimony that the cause of death was asphyxiation by compression.10 None of the experts who have re-examined the autopsy results and other relevant data, including Dr. Norton, is currently willing to say that asphyxiation by compression was the cause of death. But nor do any of the experts rule out asphyxiation by some other cause, such as suffocation,11 and none can rule out homicide as the manner of death.12 Dr. Norton remains of the opinion that the manner of death was homicide. She believes the cause of death was asphyxiation by suffocation. The only thing demonstrably “false” about Moore’s trial testimony, then, is her opinion that the death was caused by asphyxiation by compression. We cannot say for certain that her trial conclusions that the cause of death was asphyxiation, and that the manner of death was homicide, were necessarily false, or even that they were mistaken. The worst that can be said is that Moore’s trial testimony left a false impression, since most of the experts now believe that the cause of death, while it may have been some means of asphyxiation, was not asphyxiation by compression, and that the manner of death is “undetermined.”

On this state of the record, it is clear enough to me that the convicting court was *466correct to recommend that we reject the applicant’s actual innocence claim. That Moore has changed her mind that the autopsy revealed a homicide, and now believes that the manner of death is “undetermined,” does not constitute “affirmative evidence of innocence.” It would be different, perhaps, were she now willing to testify that the manner of death was something other than a homicide. Far from that, however, she persists in her opinion that the death was, at least, “suspicious.” By now declaring the manner of death to be “undetermined,” she only serves retroactively to impugn the State’s case for the applicant’s guilt; she does not affirmatively demonstrate the applicant’s innocence. No other expert has opined that the death was not a homicide. That at least one other pathologist, Dr. Norton, remains willing to characterize the death as a homicide — however assailable her opinion may be — further supports the conclusion that Moore’s re-evaluation does not affirmatively demonstrate the applicant’s innocence. After all, there is no new scientific breakthrough, no new definitive study, that has convinced Moore that her original assessment of the data was insufficiently informed from the perspective of the underlying science. Instead, she testified that she simply changed her mind, in light of certain circumstantial facts (not related to the autopsy) that she was unaware of in arriving at her original opinion,13 and with the benefit of additional professional experience since the time of the applicant’s trial.

Does Moore’s apparent mistake about the specific cause of death and her subsequent change of mind with respect to the manner of death mean that her trial testimony was “false”? If we say that it does, it will have the consequence of entitling the applicant to post-conviction habeas corpus relief even though he has not affirmatively demonstrated his innocence. The applicant concedes that there was no malfeasance on the prosecutor’s part in presenting Moore’s (apparently) honest and sincere appraisal of the autopsy data at trial.14 Should we, nevertheless, declare *467her trial testimony to be “false” for purposes of due process analysis because she has since disowned it? I believe that to do so would place too high a premium on accuracy of results as a justification for dismantling an otherwise legitimate conviction. There was nothing unfair about the trial as it occurred. It is possible, of course, to retroactively regard any given trial as “unfair” if new evidence — scientific, medical or otherwise — objectively and definitively shows that good faith trial testimony was inaccurate or false. That is the lesson 'of Estrada v. State.15 But, except with respect to her testimony as to the specific cause of death (asphyxiation by compression), proof of which is not necessary to a conviction for the result-of-conduct offense of capital murder,16 it has not objectively and definitively been established that Moore’s trial testimony was inaccurate or false. It was a medical/scientific opinion, based upon her education, training and experience-as-of-that-time. That she has since recanted it does not necessarily mean it was inaccurate or false, either now or when she expressed it at trial. Even now, she is unwilling to say that the manner of death was not homicide — only that, based on the data, she cannot currently say that it was. To call Moore’s testimony “false” under these circumstances, and thereby grant the appli*468cant habeas relief out of an overriding concern for the accuracy of the result, is essentially to grant an actual innocence claim without requiring the applicant to satisfy the usual “Herculean” burden.

With these supplemental comments, I join the Court’s opinion.

filed a dissenting opinion in which WOMACK and JOHNSON, JJ., joined.

This case raises a novel and difficult issue for the criminal-justice system.

At trial, Dr. Moore, the assistant medical examiner who performed the autopsy, testified unequivocally that the cause of seventeen-month-old Tristen’s death was asphyxiation by compression. It was a homicide. She was the State’s star witness on the child’s cause of death. Her testimony was critical to the State’s case. By finding applicant guilty of capital murder, the jury obviously believed Dr. Moore’s testimony and rejected the contrary testimony of the defense expert, Dr. Bux, who stated that the cause of Tristen’s death was “undetermined.” But Dr. Moore has now changed her mind and states that she does not know how Tristen died. She now agrees with the defense expert and believes that the cause of death is “undetermined.” Several other experienced and expert medical examiners agree with her current opinion — the cause of Tristen’s death cannot be determined by medical science. It could have been accidental asphyxiation, ie., natural suffocation because Tristen was found cold and blue in her bed with her face, including her mouth, covered by a pillowcase.1 It could have been homicidal asphyxiation by suffocation, but it probably was not homicidal asphyxiation by compression. They just do not know. These five pathologists2 admit that they cannot scientifically determine the cause of Tristen’s death. One recently retained expert, Dr. Linda Norton, a Dallas pathologist, is confident that Tristen’s death was asphyxiation by suffocation and the result of a homicidal act.3

*469In my view, this scientific uncertainty about Tristen’s cause of death raises an extremely serious concern about the accuracy of the original jury verdict. It is somewhat akin to a case in which the experts at an arson-murder trial expressed complete confidence that the fire that killed the victim was set intentionally and was the result of arson. But later, numerous other experts agree that, based on their review of the evidence and the science, no one can determine whether the fire was the result of arson or not. The cause of the fire was, in their current view, not capable of being scientifically determined as arson or accidental. It was a fire of “undetermined” and “undeterminable” origin. At trial, the experts may have been acting in perfect good faith and in accord with their current knowledge and level of expertise, but it turns out that they may have been entirely wrong. Or maybe not. Maybe it was homicide, maybe it was arson, but we do not know for certain. When scientific experts honestly and sincerely thought “X” was true at the time they testified, but the science has changed or the experts’ understanding of the science has changed and their opinions have changed, what cognizance of that change should the criminal justice system take long after a person has been convicted?

Part of the problem is that there is a fundamental disconnect between the worlds of science and of law. Science is constantly evolving by testing and modifying its prior theories, knowledge, and “truths.”4 It is a hallmark of the scientific method to challenge the status quo and to operate in an unbiased environment that encourages healthy skepticism, guards against unconscious bias, and acknowledges uncertainty and error.5 The legal system, on the other hand, “embraces the adversary process to achieve ‘truth,’ for the ultimate purpose of attaining an authoritative, final, just, and socially acceptable resolution of disputes.”6 The judicial system normally accepts that “opinions grounded in science carry their own tests for reliability and usefulness, thus inspiring special confidence in judgments based *470on them.”7 This disconnect between changing science and reliable verdicts that can stand the test of time has grown in recent years as the speed with which new science and revised scientific methodologies debunk what had formerly been thought of as reliable forensic science has increased. The potential problem of relying on today’s science in a criminal trial (especially to determine an essential element such as criminal causation or the identity of the perpetrator) is that tomorrow’s science sometimes changes and, based upon that changed science, the former verdict may look inaccurate, if not downright ludicrous. But the convicted person is still imprisoned. Given the facts viewed in the fullness of time, today’s public may reasonably perceive that the criminal justice system is sometimes unjust and inaccurate. Finality of judgment is essential in criminal cases, but so is accuracy of the result — an accurate result that will stand the test of time and changes in scientific knowledge.

Our criminal justice system does not currently have any legal doctrine, much less a constitutional doctrine, into which this situation falls comfortably. I agree with Judge Price’s concurrence that neither of our two recognized categories of “actual innocence” or “false testimony” is exactly appropriate. Dr. Moore’s change in opinion from “certain of homicide” to “undetermined cause of death” does not mean applicant is actually innocent of homicide. Nor does it mean that Dr. Moore’s trial testimony was actually “false,” at least not to the extent that she is now certain that Tristen’s death was not the result of homicide. As the State appropriately argued to the habeas judge, “It’s hard to believe that a violation of due process is established by evidence that an expert opinion may have been correct or it may have been incorrect. We’re just not sure.” Due process was not violated at the time of trial, but nevertheless, the scientific testimony that supported a finding of “homicide” in the original trial has been retracted. Dr. Moore’s current scientific uncertainty, as well as the uncertainty of four other expert pathologists, casts a pall upon the basis for the jury’s verdict and upon its accuracy.

I reiterate what I have stated before concerning the importance of ensuring accurate results in the criminal justice system.

I believe that if the criminal justice system — even when its procedures were fairly followed — reaches a patently inaccurate result which has caused an innocent person to be wrongly imprisoned for a crime he did not commit, the judicial system has an obligation to set things straight. Our criminal justice system makes two promises to its citizens: a fundamentally fair trial and an accurate result. If either of those two promises are not met, the criminal justice system itself falls into disrepute and will eventually be disregarded.8

The result in this case is not “patently inaccurate.” Yet its accuracy is clearly open to dispute. How should the habeas case be resolved when the prior verdict might have seemed accurate at the time, but everyone later recognizes that it might not have been accurate because it was based upon scientific expertise that has *471been rejected-either by the scientific community or the original scientist herself?

Given the current legitimate concerns about the scientific reliability of forensic science used in American courtrooms, I think that the criminal justice system needs some jurisprudential mechanism to deal with cases in which a prior conviction was based upon scientific evidence that has subsequently been found to be unreliable, in whole or in a specific case.9 I suspect that the Supreme Court will one day hold that a conviction later found to be based upon unreliable scientific evidence deprives the defendant of a fundamentally fair trial and violates the Due Process Clause of the Fourteenth Amendment because it raises an intolerable risk of an inaccurate verdict and undermines the integrity of our criminal justice system.

Should that rule apply in this case? Has the accuracy of the jury’s verdict been thrown into such disrepute that the courts should find a due process violation? We must not forget that, until Dr. Norton came along, the State — for good reason— joined the defense in requesting a new trial on this basis. The single most hotly contested issue in applicant’s original trial was Tristen’s cause of death:10 Did she die as the result of applicant beating her to death? Or in an asphyxiation accident (applicant suggested natural suffocation or Sudden Infant Death Syndrome), or accidental asphyxiation by compression when Tristen’s mother and a neighbor performed aggressive CPR on her? On direct appeal, the court of appeals relied upon Dr. Moore’s autopsy report and testimony to conclude that the State proved that “someone pressed the life out of Tris-ten.” 11 No expert now can conclude that *472Tristen died as a result of being “pressed” or “beaten” to death.

But even if one totally discounts Dr. Moore’s testimony and ignores the opinion of Dr. Norton,12 the evidence might well be sufficient to support applicant’s conviction for murder. At worst, the result of a finding that five out of six pathologists cannot determine the cause of Tristen’s death is only an admission that science cannot resolve the issue of whether Tris-ten’s death was the result of a homicidal act. The jurors would have to decide that crucial question based upon the rest of the evidence. In this case, the jury might have resolved the question of Tristen’s cause of death, beyond a reasonable doubt, based upon the other evidence, especially the evidence of other instances in which Tristen suffered physical injuries while in applicant’s care. As noted in my concurrence on direct appeal, the evidence showing that “Tristen repeatedly suffered various physical injuries when left in appellant’s care was admissible to prove the corpus delicti of murder.”13 The problem, however, is that we do not know whether the jury actually would have found that applicant caused Tristen’s death absent Dr. Moore’s expert scientific opinion. Who should decide whether the newly discovered unreliability of the expert scientific testimony was so crucial to the original jury’s verdict that the accuracy of that verdict can no longer be relied upon?

I fall back upon the wisdom and experience of the habeas judge — the “Johnny-on-the-Spot” factfinder to whom we will defer whenever the record supports his *473essential factual findings. This judge, who presided over both the original trial and the habeas proceeding, spent an enormous amount of time, resources, and thought in his gathering of the evidence, his review of that evidence, and his factual findings. The trial judge made twenty-two pages of findings on the following issues:

(1) Dr. Moore’s educational background and experience and her testimony at trial;

(2) Dr. Robert Bux’s educational background and experience and his testimony at trial (Applicant called Dr. Bux as a witness at trial. Dr. Bux had been the deputy chief medical examiner for Bexar County, and he testified that the cause of Tristen’s death could not be determined);

(3) the State’s reliance on Dr. Moore’s testimony at trial;

(4) Dr. Wolfs reevaluation of Dr. Moore’s conclusions;

(5) Dr. Carter’s reevaluation of Dr. Moore’s conclusions;

(6) Dr. Moore’s own reevaluation;

(7) Dr. Norton’s educational background and experience, her conclusion that Tris-ten died from asphyxia by suffocation and that her death was a homicide, and the fee she received from Montgomery County ($22,907.50);

(8) A piece of paper resembling United States currency with no identifiable markings found in Tristen’s casket;

(9) Dr. Wheeler’s educational background and experience and his conclusion that Tristen’s death was undetermined;

(10) The appointment of Mr. Milutin, an attorney who specialized in medical experts, to depose the expert witnesses;

(11) Dr. Moore’s deposition testimony that she believed the cause of Tristen’s death was undetermined and that there were other explanations for her death;

(12) Dr. Moore’s inexperience (when she performed the autopsy on Tristen she had been an associate medical examiner for just 18 months) and citations for defective and improper work;

(13) Dr. Wheeler’s deposition testimony that “he could not rule out suffocation or asphyxiation as the cause of death, but did not see any physical findings that would support any particular conclusion as to the cause of death”;

(14) Dr. Wheeler’s opinion that “the cause and/or manner of [Tristen’s] death is undetermined because the facts from the scene, from the autopsy and gross microscopic do not permit anyone to determine homicide within a reasonable degree of medical certainty”;

(15) Dr. Wolfs deposition testimony and review of the case with Dr. Luis Sanchez, the chief medical examiner for Harris County;

(16) Dr. Wolfs opinion that “he could not rule out suffocation as the cause of death, but that it was generally impossible to determine that a child’s death resulted from suffocation in the absence of a confession by the perpetrator, because of the absence of autopsy findings specific to a death by suffocation”;

(17) Dr. Wolfs testimony that the “death of a 17-month-old infant without any ascertainable cause was a rare occurrence and the death could therefore be viewed as suspicious and warranting investigation, but he could not say that it was more likely than not that foul play was involved”;

(18) Dr. Norton’s cancellation of a scheduled deposition and her leave of absence from her medical practice for six months;

(19) Dr. Norton’s inability to participate in a deposition;

*474(20) The inability of investigators to locate Dr. Norton, after a subpoena was issued ordering her to appear for a deposition;

(21) A subsequent order that the parties take Dr. Norton’s deposition at the location of her choosing; and

(22) Norton’s phone call to counsel in which she said she was under a doctor’s care, she was not physically capable of preparing for or participating in a deposition, and she did not expect to be able to participate in a deposition until January 2010, at the earliest.

The following findings are especially important to the trial judge’s ultimate conclusion:

(1) Dr. Moore’s testimony at trial was “critical” to the State’s case and “her opinions were the sole bases of the State’s case as to cause and manner of death, without which the State would not have obtained a conviction.”

(2) Dr. Moore was an agent of the State, and when she testified, she “acted in the name and for the State, was clothed with the State’s power, and her acts were those of the State.”

(3) The State did not knowingly sponsor or fail to correct false testimony.

(4) Dr. Moore “was not competent at the time of trial to offer objective and pathologically sound opinions as to cause and manner of death in this case. Her level of inexperience at the time of trial and her bias at that time toward the state are now evident. Moore’s admissions that near the time of trial she was cited for defective and improper work and was evaluated as being biased in favor of the prosecution, as well as Dr. Carter’s statements concerning the turbulence in the ME office in 1998, the concern about Moore being perceived as a witness for one particular side, and that (at that time) Moore was making the transition to the neutral position of a forensic pathologist cast grave doubt on Moore’s opinions at trial and the reasons she gave them. This is newly discovered evidence that could not have been previously discovered by applicant.”

(5) Dr. Moore’s “trial opinions as to cause and manner of death were not true. They were based on false pretenses of competence, objectivity and underlying pathological reasoning, and were not given in good faith. They were admittedly not justified by the objective facts and pathological findings in this case. Her trial opinions in this case were expert fiction calculated to attain a criminal conviction.14 This is newly discovered evidence that could not have been previously discovered by applicant.”

(6) Dr. Moore’s “giving scientifically unsupported testimony at a time when she was biased toward the State and unqualified to render opinions on the issues involved in this case equates to her giving false testimony or at least giving a false impression that was of such a nature as to contrive a conviction through the pretense of a trial which in truth was a deception of the court and jury.15 This is newly discovered evidence that could not have been previously discovered by applicant.”

(7) “While the State could not have known at trial of the falsity of Moore’s testimony, due to Moore’s public position and agency relationship with the state, her knowingly giving false testimony equates to the giving or allowing of false testimony by the state.”

*475(8) Dr. Moore’s deposition testimony is not credible. Her “recantation or reevaluation are not consistent with the facts. Moore’s claim that she did not have certain delineated facts at trial is not accurate. In addition, while she also states that her increase in training and experience give her a stronger ability to now give opinions in this case, she does not provide any specific or sound pathological reasoning to support her changed opinions.”

(9) Dr. Wolfs testimony is “especially credible and persuasive.”

(10) Dr. Carter’s original opinions were “not true,” and her new opinions lack credibility. “They are not stated with or grounded in any scientific or pathological reasoning. Her statement that she would now defer to a hypothetical ‘re-rule’ made by the HCMEO, while at the same time not giving any medical or other bases to support her opinions, is disturbing.”

(11) Although Dr. Wheeler is highly qualified, his testimony adds “very little to this case.... His evasive responses to direct questions reflected a lack of understanding of the very specific pathological issues before this Court.”

(12) Dr. Norton’s opinions are “credible and believable,” and her “extensive background and experience” are “indicative of trustworthiness.” “On the other hand, her inability to present herself for a deposition and to be subjected to cross examination lends support to the argument that her credibility and opinions as to cause and manner should be viewed as lacking the test of direct challenge. As a result, Dr. Norton’s opinions must be kept at arm’s length when being reviewed under the standards imposed on this Court in habeas proceedings.”

(13) No expert rules out asphyxia as the cause of death and they all agree that the “pathological evidence alone is not conclusive in this case, as it could support asphyxia or other causes of death.”

(14) “No expert can exclude Applicant as the perpetrator if it is a homicide, and no expert has excluded homicide as the manner of death.”

(15) “There is credible evidence and there are credible opinions that the cause of Tristen’s death is either undetermined or asphyxia by suffocation,” but no expert supports the original theory of asphyxia by compression.

(16) There is expert testimony and non-medical evidence in the trial and habeas record from which a reasonable jury could find beyond a reasonable doubt that the death of Tristen was or was not a homicide.

The habeas judge’s conclusions of law are as follows:

(1) Applicant failed to establish that he is actually innocent.

(2) Applicant “has satisfied the Schlup type standard to enter the gateway to present error that allegedly tainted the verdict at trial.”

(3) Applicant has “shown by a preponderance of the evidence, or more likely than not, that constitutional error in his trial probably caused a verdict in which there can be little confidence and which is fundamentally unfair.”

(4) Applicant established by a preponderance of the evidence that his due process and due course of law rights were violated.

(5) Applicant also established that he did not receive a fair trial before an impartial jury as guaranteed by the Sixth Amendment of the U.S. Constitution and Article 1, § 10 of the Texas Constitution.

I certainly agree with the first legal conclusion: applicant has not established his *476actual innocence — not even close. But, given the experienced trial and habeas judge’s legitimate and serious concerns about the impact of Dr. Moore’s expert testimony at trial on the critical and hotly disputed issue of Tristen’s cause of death, I agree that applicant did not receive a fundamentally fair trial based upon reliable scientific evidence (or the honest admission that science cannot resolve that critical issue). Despite every participant’s honesty and good faith,16 this is a case that should be retried to ensure the accuracy of our verdicts and the integrity of our system.

I therefore most respectfully dissent to this Court’s rejection of the trial and habe-as judge’s factual findings and recommendation that we grant applicant relief and remand this case for a new trial.

filed a dissenting opinion.

I respectfully dissent. I conclude that Neal Hampton Robbins is entitled to relief on his application for a writ of habeas corpus on the ground that he was denied due process of law by the State’s use of false testimony to obtain his conviction.1

I. False Testimony

The Due Process Clause of the Fourteenth Amendment is violated when the State knowingly or unknowingly uses perjured testimony to obtain a conviction. Ex parte Chabot, 300 S.W.3d 768, 770-71 (Tex.Crim.App. 2009); Ex parte Napper, 322 S.W.3d 202, 242 (Tex.Crim.App. 2010) (“Chabot simply stands for the proposition that the preponderance of the evidence standard is appropriate for the unknowing use of perjured testimony that the habeas applicant had no prior opportunity to discover.”). The term “perjury” in this context requires proof that the testimony “gives the trier of fact a false impression,” but it does not require proof of the elements of “perjury” as that term is defined in the penal code. See Ex parte Ghahremani, 332 S.W.3d 470, 477-78 (Tex.Crim.App. 2011); see also Napper, 322 S.W.3d at 242 (citing Estrada v. State, 313 S.W.3d *477274, 287 (Tex.Crim.App. 2010)) (“[W]e held on direct appeal that false testimony that was not perjury resulted in a due process violation when there was ‘a fair probability that [the] death sentence was based upon ... incorrect testimony.’ ”).

The trial court found that Dr. Moore’s trial testimony was false. The trial court’s findings state that “Dr. Moore’s trial opinions were not true. They were based on false pretenses of competence, objectivity, and underlying pathological reasoning, and were not given in good faith.” The trial court characterized her testimony as “expert fiction calculated to attain a criminal conviction.” The record supports the trial court’s characterization concerning the falseness of the testimony.

The record shows that, as the sole witness establishing cause and manner of death for the State at Robbins’s trial, Dr. Moore testified that, based on her scientific opinion beyond a reasonable doubt, the cause of Tristen Rivet’s death was asphyxia due to compression of the chest and abdomen, and the manner of death was homicide. In her evidence concerning this application for a writ of habeas corpus, she now concludes that the cause of death was, beyond a reasonable doubt, not compression asphyxia, and undeterminable as to homicide, asphyxial or otherwise. Dr. Moore’s subsequent testimony is a complete refutation of her trial testimony because, although her trial testimony stated that, beyond a reasonable doubt, the cause of death was compression asphyxia and the manner of death was homicide, she now says that the cause and manner of death are, beyond a reasonable doubt, “undeter-minable.” Both positions cannot be true. This wholesale refutation of her previously professed scientific certainty nullifies the veracity of the conclusion itself.

I recognize that, as noted by the majority opinion, the record shows that neither Dr. Moore nor any of the other testifying experts can “exclude” asphyxial homicide as a possible cause of death or “rul[e] out other reasonable hypotheses by which Tristen died.” In other words, because Dr. Moore presently acknowledges that the cause of Tristen’s death could possibly have been homicide and possibly by asphyxiation, the majority opinion determines that her new testimony does not show that her earlier testimony is false. But Dr. Moore is merely acknowledging the possibility that this cause and manner of death could be true because her opinion is, beyond a reasonable doubt, that she does not know the cause and manner of death. The fact that a witness acknowledges a mere possibility of an alternative hypothesis is not a failsafe escape for due process violations.

The Supreme Court has disallowed this technical splicing of the truth to avoid due process violations. In evaluating whether evidence is false, it has focused on whether the testimony, taken as a whole, gives the jury a false impression. See Alcorta v. Texas, 355 U.S. 28, 31, 78 S.Ct. 103, 2 L.Ed.2d 9 (1957) (holding Alcorta’s due process rights violated when the State, although not soliciting false evidence, allowed it to go uncorrected when it appeared, and stating, “It cannot seriously be disputed that [the witness’s] testimony, taken as a whole, gave the jury the false impression that his relationship with petitioner’s wife was nothing more than that of casual friendship.”).

Furthermore, the Supreme Court has determined that a statement is false even when the witness has acknowledged the existence of a similar situation. See Napue v. Illinois, 360 U.S. 264, 270, 79 S.Ct. 1173, 3 L.Ed.2d 1217 (1959) (holding Na-pue’s due process rights violated because witness’s testimony that public defender had offered to help him if he testified for *478the State did not turn “what was otherwise a tainted trial into a fair one” when true evidence showed that prosecutor had promised “consideration” if witness testified at Napue’s trial). These Supreme Court decisions illustrate that a due process violation cannot be excused by the mere acknowledgment that it is possible that the trial testimony could have been true. Dr. Moore’s new testimony, when taken as a whole, is that, beyond a reasonable doubt, the manner and cause of death is undeterminable, and it is that testimony that was improperly precluded from Robbins’s trial.2

Perhaps Dr. Moore’s testimony could not be called “false” if, for example, she consistently determined, beyond a reasonable doubt, that the manner and cause of death could be established with scientific certainty and that the manner of death was homicide, but was uncertain whether the cause of death was asphyxia by some means other than compression. See Berger v. United States, 295 U.S. 78, 82, 55 S.Ct. 629, 79 L.Ed. 1314 (1935) (“The true inquiry, therefore, is not whether there has been a variance in proof, but whether there has been such a variance as to ‘affect the substantial rights’ of the accused.”). But Dr. Moore’s changed testimony is not merely a variance in proof. Her present position acknowledges that the cause and manner of death could possibly be natural causes or homicide and that both are equally likely. An acknowledgment that trial testimony could possibly be correct because no one can determine the cause and manner of death with scientific certainty is vastly different from evidence that the cause and manner of death are proven beyond a reasonable doubt with scientific certainty. I, therefore, agree with the trial court’s assessment that the record shows that Dr. Moore’s testimony was false.

II. Harm Analysis

Having determined that Dr. Moore’s false testimony constituted “perjury” in the context of this due process analysis, I conclude, as did the trial court, that Robbins has shown by a preponderance of the evidence that the use of this false evidence contributed to his conviction. It is undisputed that Robbins had no opportunity to discover Dr. Moore’s bias and improper intentions prior to her recantation, and, therefore, resolution of this case rests on whether he met his burden to show harm. Regardless whether the State’s use of the perjury was knowing3 or unknowing, the *479standard of harm requires that the applicant prove by a preponderance of the evidence that the error contributed to his conviction or punishment. See Ex parte Fierro, 934 S.W.2d 370, 374-75 (Tex.Crim.App. 1996) (“[W]e hold that the knowing use of perjured testimony is trial error, subject to the harmless error standard applicable on habeas corpus,” and “we hold that the applicant has the burden to prove by a preponderance of the evidence that the error contributed to his conviction or punishment.”); Chabot, 300 S.W.3d at 771 (“Although the present case involves unknowing, rather than knowing, use of testimony, we see no reason for subjecting the two types of error to different standards of harm.”); Napper, 322 S.W.3d at 244 (“For habeas corpus proceedings involving the unknowing use of perjured testimony, the harm standard would be preponderance of the evidence”).

The record supports the trial court’s findings that Dr. Moore’s testimony was “critical” to resolution of the case and that “her opinions were the sole bases of the State’s case as to cause and manner of death, without which the State would not have obtained a conviction.” We must defer to these findings, as they are amply supported by the record. See Ghahremani, 332 S.W.3d at 478 (requiring deference to findings supported by record). Alone, Dr. Moore conducted the autopsy of Tris-ten’s body and was thus the sole testifying witness with personal knowledge of the medical findings. And her trial testimony alone professes that the cause of Tristen’s death was, beyond a reasonable doubt, homicide (asphyxial or otherwise). Absent this trial testimony, there is no evidence of the cause of death — that is, no evidence that a crime occurred — because Robbins’s defense expert medical examiner, Dr. Robert Bux, testified that Tristen’s death could have been from natural causes. Dr. Moore now agrees with Dr. Bux’s medical assessment. Without Dr. Moore’s trial testimony, the jury would have been left solely with the evidence from Dr. Bux, who stated that the cause and manner of Tris-ten’s death are undeterminable.

Mere consideration of whether the jury would have convicted based solely on the remaining circumstantial evidence absent Dr. Moore’s testimony does not sufficiently measure her testimony’s materiality. See United States v. Johnson, 149 F.2d 31 (7th Cir. 1945) (“ ‘There is no way for a court to determine that the perjured testimony did not have controlling weight with the jury, and, notwithstanding the perjured testimony was contradicted at the trial, a new light is thrown on it by the admission that it was false; so that, on a new trial, there would be a strong circumstance in favor of the losing party that did not exist and *480therefore could not have been shown, at the time of the original trial.’”) (quoting Martin v. United States, 17 F.2d 973, 976 (5th Cir. 1927)). First, Dr. Moore’s subsequent testimony actually corroborated defense witness Bux’s testimony that the cause of death was undeterminable. See Alcorta, 355 U.S. at 31-32, 78 S.Ct. 103 (in reversing judgment, stating, “If Castille-ja’s relationship with petitioner’s wife had been truthfully portrayed to the jury, it would have, apart from impeaching his credibility, tended to corroborate petitioner’s contention that he had found his wife embracing Castilleja.”). Second, her testimony that was introduced at trial affirmatively misled jurors to reach a damning conclusion on the most crucial issue that, but for that testimony, they likely would not have reached. See id. at 32, 78 S.Ct. 103 (determining that applicant’s due process rights were violated by misleading testimony denying sexual relationship between deceased and witness that was pertinent to proof of sudden passion arising from an adequate cause). Dr. Moore’s trial testimony thus prevented “a trial that could in any real sense be termed fair.” See Napue, 360 U.S. at 270, 79 S.Ct. 1173.4

III. Conclusion

I conclude that Robbins’s due process rights were violated by the false, material trial testimony by the State’s sole medical expert establishing cause and manner of death. In accordance with the recommendation by the convicting court to grant relief from the judgment, I would remand for a new trial.