OPINION
Newell, J.,delivered the opinion of the Court
in which Johnson, and Richardson, JJ., joined.According to Applicants’ expert, Dr. Alexandria Doyle, the sexual-assault allegations in this case do not pass “the smell test.” This emotional response certainly captures the sense of outrage that so many harbor about these cases. Whether it is in articles or a documentary, these cases involving “The San Antonio Four” have been well dissected in popular media. See e.g. Southwest of Salem: The Story of the San Antonio Four (Deborah S. Esquenazi Productions 2016); Bridgette Dunlap, Inside Case Behind Wrongful Conviction Doc ‘Southwest of Salem’, Rolling Stone, Oct. 13,2016; Maurice Chammah, Case of “San Antonio Four” Set to Enter its Final Act, The Texas Tribune, March 29, 2015; Mau*288rice Chammah, A Growing Battle for Exoneration, N.Y. times, Nov. 18, 2012.
But we are not asked to apply a “smell test.” Rather, we are asked to decide whether the newly available evidence of innocence undermines the legally sufficient, but hard-to-believe version of events that led to the convictions of these four women. We hold that it does and that these four women have unquestionably established that they are innocent of these charges.
I. Introduction
In the summer of 1994, two young girls alleged that four young lesbian women, including the girls’ aunt, had spontaneously and violently gang-raped them on two occasions within a single week. The who-what-when-and-where changed from the outcries, to the statements made to the police, to the statements made to the examining doctor, to the testimony at two trials. But those inconsistencies were easy to set aside given the physical findings associated with child sexual abuse found by Dr. Nancy Kellogg, who asserted that the older child showed physical, objective signs of sexual abuse: In light of Dr. Kellogg’s testimony, the girls’ stories had the ring of truth.
These inconsistencies can no longer be set aside in light of what we know now. Dr. Kellogg has retracted her testimony about the physical indicators of past trauma. She now agrees with the defense that there are no definitive signs of sexual abuse, and she has acknowledged that her testimony at trial was wrong. All parties and courts, including this one, agree that all four Applicants are entitled to have their convictions and sentences vacated because of the introduction of what is now known to be scientifically invalid or inaccurate evidence.
But there is a great deal more that casts doubt upon the reliability of the convictions in these cases than just the unreliable scientific evidence. One of the complainants, the younger sister, now an adult, has recanted her testimony and explained how and why she and her sister made up a story about her aunt and her aunt’s three friends. Expert testimony regarding false allegations of sexual assault now establishes that this complainant’s recantation was genuine, voluntary, and sincere, and her story fits the profile of other false claims of sexual abuse. Though the other complainant has not recanted her trial testimony, her sister’s recantation, credited by the trial court, cannot be logically reconciled with the remaining testimony establishing guilt.
Furthermore, the Applicants have also presented evidence that the complainants’ father, Javier Limón, has engaged in a pattern of threatening behavior towards the complainants and false allegations of sexual assault to gain leverage in disputes over custody of his children. The Applicants have presented credible testimony that the complainants’ father threatened and assaulted the two complainants to ensure that they accused these women of sexual abuse. Moreover, the Applicants have presented credible expert testimony explaining how the techniques used during the investigation of the alleged crimes could have reinforced the complainants’ childhood belief in a crime that never occurred.
Finally, the Applicants have presented new expert testimony that they are not sex offenders, None of the four Applicants fit the profile for sex offenders, and psychological evaluations have confirmed this. From the moment these allegations were made, all four Applicants have consistently maintained their innocence and each other’s innocence despite multiple, separate interviews.
*289We conclude that now, with this clear and convincing evidence establishing innocence combined with the lack of reliable forensic opinion testimony corroborating the fantastical allegations in this case, no rational juror could find any of the four Applicants guilty of any of the charges beyond a reasonable doubt. We agree with the habeas court that relief is required based on new scientific evidence, but we also hold that the Applicants have carried them burden to establish a claim of actual innocence. Although the habeas court did not recommend granting relief on actual-innocence grounds, it did so strictly because only one of the two complainants recanted. It relied upon a “legal sufficiency5’ analysis without considering the overwhelming evidence of innocence and its impact upon the State’s already weak cases. We disagree with the habeas court’s apparent assessment that the lack of a recantation from one of the two complainants is fatal to Applicants’ actual-innocence case. We exercise our authority to reach the contrary conclusion, and, accordingly, grant relief under a more comprehensive and robust actual-innocence analysis. See Ex parte Reed, 271 S.W.3d 698, 727-28 (Tex. Crim. App. 2008).
II. The Facts as Presented in the Two Trials
In the summer of 1994, nine-year-old V.L. and her seven-year-old sister, S.L., stayed with their nineteen-year-old aunt, Elizabeth Ramirez, for several days while their mother, Rosemary Camarillo, was in Colorado.1 During the course of their stay with Elizabeth, who shared the apartment with her one-time girlfriend, Kristie May-hugh, another couple, Anna Vasquez and Cassandra Rivera, visited the apartment frequently. Cassandra and her children even spent the night a few times during the week of the girls’ visit.
The girls’ grandmother, Serafina Li-món,2 said that, when the girls came home, they were not acting “normal;” they were subdued, scared, and refused to make eye contact. In mid-September, Serafina noticed the girls playing with their dolls in a sexual manner. When she asked the girls why they were doing this, V.L. told Serafi-na that she and her sister had been sexually assaulted at their aunt’s apartment by the four women.
The girls’ basic accounts were this: the four women stripped them, held them down, fondled them, and stuck objects and liquids into their vaginas. The girls stated they were assaulted on two different occasions. The assaults were back-to-back the first time, V.L. first and then S.L., and then simultaneously the next time. According to both girls, these assaults were completely spontaneous without any suspicious behavior leading up to them.
By V.L.’s account, the assaults occurred a couple of days into the visit to their aunt’s apartment. V.L. and her sister were playing outside when both girls were called inside. According to V.L., the women began yelling things at her like “why did you do this? Why did you do that?” V.L. was inconsistent when explaining where S.L. was during the first assault; S.L. was either locked outside, or in the living room. V.L. was brought into Elizabeth’s bedroom, where, according to V.L., *290Elizabeth held her down while the other women started touching her; the four women did not touch each other, only her. V.L. said she kicked and screamed as the women put stuff inside her vagina — liquid stuff, a powder, and “a tampon or something.” She said that it hurt. And after they finished, they told her to go take a shower. According to V.L., she heard S.L. screaming and crying in the bedroom as she came out of the shower. Then, S.L. came out of the bedroom with no pants on, and the women told S.L. to take a shower. V.L. then went back outside to play.
V.L. said that the next time she was assaulted, it was again in the bedroom by Anna and Liz (and maybe Cassie), while S.L. was assaulted in the living room by Cassie. Or, they were assaulted together. She made a statement that “they started putting some kind of stuff in us. Then they put liquid in us. Cassie was the one that put the liquid in me. They did this to S.L., too.”
S.L.’s account was similar with regard to the fundamentals of being held down and having things inserted into her vagina. But S.L. said that, during the first assault, when she came inside, she heard V.L. screaming in the bedroom. She tried to peek in. Then everyone came out of the bedroom, and she asked V.L. if she was okay. According to S.L., V.L. went outside to play, and the four women took her pants and underwear off and laid her down on the floor of the living room. Then, Cassie “put something in my private” while the other women were “holding me down.” S.L. indicated that the second assault happened the next day in the living room, when “they put the same thing in my private.” Both girls said that they were threatened by the women though they were unclear about whether it was with a knife, a gun, or two guns, and whether it was by Liz, Anna, Liz and Anna, or Cassie and told to keep quiet about the assaults.
Discerning a coherent picture of the alleged assaults from the different versions provided by each complainant takes considerable intellectual effort. There are multiple different versions of events, each one differing from the others, sometimes irreconcilably so. And there are significant inconsistencies between the versions told by each complainant. While there appears to have been enough consistency to barely cover the essential elements of the offenses at issue, the stories by themselves provided weak evidence of guilt at best.
On September 28th, the girls were taken by their father, Javier Limón, to a clinic for sexual-assault exams. Dr. Kellogg, an expert in the science of physical findings associated with child sex abuse, took histories from the girls and examined them. She made physical findings that were consistent with the history of sexual abuse each girl related.
Meanwhile the four women maintained their complete innocence. None had ever been accused of any kind of criminal, violent, or otherwise anti-social behavior, toward children or anyone else. Each completely cooperated with the homicide detective assigned as the lead investigator in the case, without invoking their right to an attorney. All submitted to individual interrogations and gave signed written statements, expressing bewilderment at the accusations and maintaining their innocence.
Elizabeth Ramirez was tried first and alone, in 1997, for sexually assaulting V.L. V.L. testified, but S.L. did not. Dr. Kellogg testified that V.L.’s statements about the assaults were “spontaneous ... detailed ... guileless ... uncontrived.”3
*291Dr. Kellogg said V.L.’s exam was not normal: There was a scar on the hymen, a healed tear. The scar indicates “painful” penetration. And although there was no way to tell when the penetration happened, it was no longer than nine years before the exam because V.L. was nine at the time of the exam. Dr. Kellogg also testified that playing with dolls in a sexual manner was acting out and “with sexual acting out it tends to be more specifically linked to sexual abuse.”
Elizabeth, age 20, testified in her own defense. She said she did not do anything to the children, she had “no knowledge how they could even think of me doing something like that.” She said there was no gun, no knife, no threat. Nevertheless, she was convicted of aggravated sexual assault of a child and indecency with a child, and sentenced to 87 and 15 years’ imprisonment. The court of appeals affirmed. Ramirez v. State, 04-97-00144-CR, 1998 WL 412487 (Tex. App.-San Antonio July 22, 1998, pet. ref'd) (not designated for publication).
Kristie Mayhugh, Cassandra Rivera, and Anna Vasquez were tried together a year later for sexually assaulting both S.L. and V.L. Both girls testified. Between the two trials, Dr. Kellogg found out that the girls had undergone sexual assault exams in 1992.4
Dr. Kellogg testified that, in relating the events, S.L. was scared, disconnected, un-contrived, and very open. Dr. Kellogg said that, when she examined S.L.’s genitals, she had a lot of redness, which could be caused by irritation, inflammation, infection, or sexual trauma. Also, S.L.’s hymen was thickened, which “could be due to abuse with trauma,” or could have “nothing to do with trauma but it’s a normal variation.” She observed that, in the 1992 exam, S.L.’s hymen was thickened, and red, but not as red as it was in the 1994 exam. On cross examination, she acknowledged that S.L.’s exams were normal but “it doesn’t rule out abuse.”
Dr. Kellogg again testified about V.L.’s statement and abnormal exam — with the scarring on the hymen indicating vaginal penetration. And because the scar was not present in the photos taken during the 1992 exam, the injury causing the scar happened “since the exam of 92.”
All three defendants testified. All three denied any inappropriate contact with the girls. All three were convicted of two counts of aggravated sexual assault of a child and two counts of indecency with a child. All three were sentenced to fifteen years’ imprisonment on the aggravated-sexual-assault charges, and ten years’ imprisonment on the indecency charges. The court of appeals in three separate, but nearly identical opinions, affirmed. May*292hugh v. State, Nos. 04-98-00262-CR & 04-98-00263-CR, 1999 WL 1246925 (Tex. App.-San Antonio Dec. 22, 1999, pet. ref'd); Rivera v. State, Nos. 04-98-00186-CR & 04-98-00187-CR, 1999 WL 1246934 (Tex. App.-San Antonio Dec. 22, 1999, pet. ref'd); Vasquez v. State, Nos. 04-98-00245-CR, 04-98-00246-CR, 1999 WL 1246932 (Tex. App.-San Antonio Dec. 22, 1999, pet. ref'd).
III. Habeas Proceedings Part 1: The 11.073 Claim
All four women filed identical applications for post-conviction relief on the basis of new science and actual innocence. Judge Mary Roman heard the Article 11.073 new-science claim. This claim was based on Dr. Kellogg’s recantation of the core of her trial testimony: that V.L.’s scarred hymen indicated penetration.
The State and Applicants submitted agreed findings and a conclusion that Applicants were entitled to Article 11.073 relief based on having met the Article 11.073 standard. The findings recognized that new scientific studies within the field of pediatrics — showing that while injured hymens do heal, they do not leave scars in pubertal and prepubertal girls — contradicted the medical testimony presented at the trials. Under the current scientific knowledge, and upon review of the original photographs taken during the sexual assault examinations, there were, in fact, no physical signs of abuse. The findings on the joint trial cases, were, in part
• Dr. Kellogg recognizes that if the medical science in this area (as presented in Dr. McCann’s study published in 2007) had been available to her in 1997 or in 1998, as an expert in the field of child sexual abuse she “would not have testified that the finding was indicative of trauma to the hymen.”
• The only scientific evidence before the jury that a sexual assault occurred was Dr. Kellogg’s original trial testimony that she observed a “scar” on V.L.’s hymen that was the result of “a tear that had healed.”5
• Dr. Kellogg testified that this medical finding was consistent with the sexual abuse V.L. described and “vaginal penetration” with some unknown object.
• At trial, the State relied heavily on Dr. Kellogg’s expert testimony to establish first, that the crime as alleged, in fact occurred, and second, to corroborate the complaining witness’s testimony.
• Dr. Kellogg’s scientific testimony of V.L.’s “hymenal scar,” her physical findings and her expert opinions and conclusions derived from her physical findings, corroborated V.L.’s trial testimony that she was sexually abused by Applicant^].
• Although Dr. Kellogg did not testify to finding a “hymenal scar” during S.L.’s sexual assault exam, S.L. and V.L.’s testimonies were so inextricably intertwined that Dr. Kellogg’s scientific testimony undoubtedly corroborated S.L.’s testimony. Due to this connection, it is also reasonable to believe that *293Dr. Kellogg’s scientific testimony factored into the jury’s evaluation of S.L.’s sexual assault exam, albeit deemed “normal.”
• There is a reasonable probability that the outdated medical testimony concerning the “hymenal scar” was indicative of penetrating trauma to the hymen contributed to the jurors’ belief that the offenses did occur.
The agreed conclusion was that “more likely than not had this newly available, relevant scientific evidence regarding hymenal injuries been presented at trial, Applicant[s] would not have been convicted of the two counts in the indietment[s].”
Nearly identical agreed findings and conclusions were made in Elizabeth Ramirez’s case. Judge Roman accepted and signed the findings and conclusions, recommending that relief be granted on Applicants’ Article 11.073 claims. These findings are supported by the record, and we hereby adopt them. We, therefore, grant habeas corpus relief on Applicants’ claims that, more likely than not, they would not have been convicted had the newly available scientific testimony been presented to the jury. The remaining claims of actual innocence were referred to Judge Pat Priest, who had presided over the joint trial.6
IY. Habeas Proceedings Part 2: The Actual-Innocence Claim
A two-day evidentiary hearing was held on the Applicants’ actual-innocence claims. At this hearing, Rosemary Camarillo testified about her difficulties with her ex-husband, Javier, regarding their children. According to Rosemary, Javier pulled a gun on her and threatened to kill her during an argument after she and her kids had moved to Colorado. All of the children saw the incident. During that same weekend, Javier took their three children back to San Antonio over Rosemary’s objections. That incident led to a custody battle.
To gain leverage in the custody dispute, Javier made false accusations that a man in Colorado, whom the girls had never been left alone with, had sexually assaulted the girls. Javier also made a false accusation that a non-existent 10-year-old boy in Colorado had sexually assaulted S.L. That accusation led to a police report and the examinations of the girls in 1992, but the investigation was ultimately dropped. According to Rosemary, Javier and his mother, Serafina, had also made accusations against others about sexually assaulting children from Javier’s other relationships. Those allegations were also false; one turned out to be based on diaper rash.
Rosemary stated that she has always believed her sister to be completely innocent of these charges and has never known her sister to possess or talk about having a gun. She testified that S.L. confirmed that when she came to live with her as an adult, telling her, “I know that nothing happened, you know, with my Aunt Liz and her friends .... all the memories that I have is memories of good things that we did together, where she would take us shopping, she would take us to eat, her and her friends. We would go to the park. They would take us swimming.... I don’t remember anything bad happening to us.”
S.L., twenty-seven years old at the time of the hearing, fully and completely recanted her claims of sexual abuse and explained how the false allegations originated, how they evolved, and how they were encouraged. She explained that she remembered her mom and dad fighting over her and her sister and her brother, Max, when they were in Colorado, and that the *294“cops came out.” She could remember visiting the apartment of her Aunt Liz with her sister when they were small and that they liked going there to visit. She said that the allegations against her aunt and the other women arose when her grandmother saw her, her sister V.L., and her female cousin playing a game.
As S.L. put it, “Me and my sister and cousin were in the room playing, and I was pretending to drive a car. And my sister V.L. and my cousin were in the back” pretending to be mom and dad and they were kissing, “using tongues.” Their grandmother saw them and started screaming at them. Javier was called. According to S.L., they were actually mimicking behavior they had seen their father engage in with other women in front of them. But when she told her father this, he insisted that she learned it from watching her lesbian aunt.
S.L. testified at the habeas hearing that she heard Javier yell at V.L., saying, “You know your aunt did something to you, you know your Aunt Liz did something to you, you just need to be honest, what did she do, who showed you how to kiss, who showed you how to make out like that.” S.L. stated that they were afraid of going against Javier because if they did so, he would hit them.
She explained that she was never in the apartment alone with all of these Applicants. She said that the first time she ever saw Anna Vasquez was when a police officer “pulled out a book, and asked us to pick out one of the ladies that touched us. And we — I’m going to say T,’ pointed out the wrong lady. And then he tells me no, and shows me the picture.”
She stated that, from the time Javier told them that something happened to them through the medical examination and the police report, she and V.L. were coached on what to say. If they got the story wrong, Javier would strike them. And after the trials, Javier never allowed them to talk about their aunt being in prison. “One time I had asked him about it. And I said, what exactly happened, and why is it they’re there. And he told me to shut up and never bring it up again.”
According to S.L., Javier was an abusive father, and when she was fourteen, she tried to commit suicide. She was put in treatment. After a few weeks she moved to a group home. She never moved back in with Javier. At nineteen, she told her counselor that she had never been sexually assaulted, and her counselor urged her to do the right thing and come forward. She said that she had hesitated because she was concerned about getting in trouble for not having told the truth in court. S.L. talked to Javier on the phone to tell him that “I was going to come and speak to somebody about the case, because I don’t remember it happening .... And when I did tell him that, he told me if I did that, that he was going to hit me where it hurt.” That same year, 2012, he accused her of being a bad mother and attempted to use the courts to take her children away, though his effort failed and she ultimately kept her children.
Finally, S.L. addressed her sister’s claims of abuse. According to S.L., when she was about thirteen, she talked to V.L. about why their Aunt Liz was in prison. “I told her, I don’t think anything ever happened to us. And she said it may not have happened to you but it happened to me.” She last talked to V.L. about three years prior to the hearing. V.L. did not testify at the hearing.
At the hearing, Applicants also presented testimony from Maria Molett, the executive director of the Counseling Institute of Texas. Molett performed psychosexual evaluations on Elizabeth, Cassandra, and *295Kristie, and looked at the records of the person who did the evaluation on Anna. She concluded that she never would have accepted the women for sex-offender treatment because “these people are not sex offenders” and so there would be nothing to treat.7
All four Applicants testified at the hearing — as they had at their respective trials — that they are innocent of the charges. Applicants also introduced testimony from forensic psychologist, Dr. Alexandria Doyle, whose expertise is in evaluating claims of sexual assault and recantations of such claims. She opined that the original claims were fantastic — nothing similar to what is typically seen in true child sexual-abuse cases — but that the recantation was credible.
At the end of the hearing, Applicants argued to the habeas court that, given all the information now known and the extremely weak and unreliable nature of the only evidence left to support the convictions, no reasonable juror could find Applicants guilty beyond a reasonable doubt. The State did very little throughout the whole hearing, asking very few questions on cross-examination and declining to put on any evidence. Significantly, the State did not recommend denying or granting relief on actual innocence, but rather stated that, now that it is known that Dr. Kellogg’s testimony in the trials in the 1990’s was wrong, what is left is “purely the credibility of the witnesses, which is for the Court to determine.”
IY. A. The Herrera Standard
This Court recognizes two types of “innocence” claims. The one at issue in this case — a Herrera claim — is a substantive claim in which the person asserts a bare claim of innocence based solely on newly discovered evidence. Ex parte Elizondo, 947 S.W.2d 202, 205 (Tex. Crim. App. 1996); Ex parte Brown, 205 S.W.3d 538, 544 (Tex. Crim. App. 2006).
Herrera claims are evaluated on the assumption that the trial that resulted in conviction had been error-free. In such a case, when a petitioner has been tried before a jury of his peers, with the full panoply of protections that our Constitution affords criminal defendants, it is appropriate to apply an extraordinarily high standard of review. Ex parte Franklin, 72 S.W.3d 671, 676 (Tex. Crim. App. 2002). Thus, to succeed in an actual-innocence claim, the applicant must show by clear and convincing evidence that, despite the evidence of guilt that supports the conviction, no reasonable juror could have found the applicant guilty in light of the new evidence. Brown, 205 S.W.3d at 545; Ex parte Tuley, 109 S.W.3d 388, 392 (Tex. Crim. App. 2002).
An applicant must also prove that the evidence he relies on was not known to him at the time of trial and could not be known to him even with the exercise of due diligence. Brown, id. Although many actual-innocence cases are based on a single piece of new evidence such as DNA or the recantation of a victim or witness, we have made clear that “multiple pieces of newly discovered evidence” can together make a meritorious case for relief. Ex Parte Miles, 359 S.W.3d 647, 671 (Tex. Crim. App. 2012). In practice, we have highlighted certain pieces of new evidence and discussed whether the new evidence persuasively establishes innocence when comparing it to the evidence establishing guilt. See e.g. Ex Parte Navarijo, 433 *296S.W.3d 558, 568 (Tex. Crim. App. 2014) (highlighting pieces of evidence and evaluating their persuasiveness before denying an actual-innocence claim). Ultimately, we look to whether the totality of the new evidence of innocence unquestionably establishes that a jury would not have found the defendant guilty in light of the new evidence when weighed against the old evidence establishing guilt. Brown, 205 S.W.3d at 545
The evidence presented in support of a Herrera innocence claim must be “affirmative.” Franklin, 72 S.W.3d at 678. Once the applicant provides such evidence, it is appropriate for the habeas court to proceed with the weighing of new evidence tending to show innocence against the evidence of guilt produced at trial. The habe-as court then makes findings of fact and conclusions of law, and a recommendation to this Court.
While we generally defer to findings of fact when the trial court is in a better position to determine witness credibility, we nevertheless can exercise our authority to make contrary or alternative findings and conclusions when our independent review of the record reveals that the trial judge’s findings and conclusions are not supported by the record. Ex parte Reed, 271 S.W.3d 698, 727 (Tex. Crim. App. 2008); see also Ex parte Weinstein, 421 S.W.3d 656, 664 (Tex. Crim. App. 2014) (citing Ex parte Chavez, 371 S.W.3d 200, 207 (Tex. Crim. App. 2012)), Ex parte Flores, 387 S.W.3d 626, 634-35 (Tex. Crim. App. 2012). Significantly, we apply a de novo standard of review to the legal question, “Does the applicant’s new evidence, viewed in the light most favorable to the habeas court’s factual findings and credibility determinations, actually prove, by clear and convincing evidence, that a jury would acquit him?” See Rivera v. State, 89 S.W.3d 55, 59 (Tex. Crim. App. 2002). In answering this question, we consider the habeas court’s conclusions of law as well as his recommendation, but it is nonetheless a legal conclusion that this Court, as the final decision maker in habeas applications, must make. Reed, 271 S.W.3d at 727.
Although courts must carefully examine claims of actual innocence — even one made many years after the alleged crime — recantations in sexual assault cases are not rare. Such post-conviction claims should not be accepted without close scrutiny nor, generally, without strong corroboration by independent evidence. Ex parte Thompson, 153 S.W.3d 416, 420-21 (Tex. Crim. App. 2005) (granting relief on the basis of actual innocence where complainant’s recantation was corroborated by normal medical examination results and evidence of manipulated allegations of abuse). For instance, even in a case where the trial court found a complaining witness’s recantation credible we have nevertheless denied relief to an applicant convicted of aggravated sexual assault of a child because of “(1) the lack of detail in the complainant’s recantation testimony at the habeas hearing and the jury’s rejection of evidence of the complainant’s pre-trial recantation, and (2) the existence of inculpa-tory medical testimony that has not been otherwise explained by the complainant’s recantation.” Navarijo, 433 S.W.3d at 568. To support a finding of actual innocence, a recantation must be direct, specific and certain. Franklin, 72 S.W.3d at 678; Brown, 205 S.W.3d at 547.
Y. Applicants Presented Substantial Credible and Persuasive Evidence Establishing Their Innocence
This Court agrees with Applicants, the State, and Judges Roman and Priest that the newly available scientific evidence establishes, by a preponderance of the evidence that Applicants would not have been *297convicted. The question remaining is whether the additional, newly available evidence when compared to the evidence establishing guilt moved this case to the next level, that is, where no rational juror would have convicted Applicants. The habeas court made several findings that suggest he found S.L.’s, Elizabeth’s and Rosemary’s testimony at the habeas hearing credible, as well as the testimony from the expert that none of the Applicants fit the profile of a sexual offender — including the following:
• During the time that Elizabeth Ramirez lived with the Limón family, a very bad relationship developed between Elizabeth Ramirez and Javier Limón. Elizabeth Ramirez testified at the writ hearing that Javier made sexual advances toward her that she rejected, and when she became pregnant by another man, Javier wanted to serve as father figure for the child, which she also rejected. Elizabeth Ramirez did not tell Rosemary Limón about Javier’s advances.
• Since the trial, Javier and Rosemary have divorced. Since the divorce, Javier has made false allegations that a Mr. Aguirre in Colorado, as well as a ten-year-old boy in that state, had sexually assaulted these two complainants. Their mother lost contact with these complainants because Javier was “difficult with the girls,” who told their mother they wanted nothing to do with him. This is what occasioned the girls being in Colorado with their mother, but he went there, pulled a gun on their mother (who took it away from him), and brought the girls back to San Antonio via Greyhound bus.
• S.L. asserts that Javier was abusive to all the children and that it was Javier who instigated these complaints. All of the Applicants have expressed that their sexual orientation is toward members of the same sex, and when Javier saw the complainants and a female cousin kissing each other, he assumed that they had done that due to the influence of “Aunt Liz” (Elizabeth Ramirez) and her friends (the other Applicants). Javier and his mother, Serafina talked to the complainants, suggesting that the girls had seen improprieties at Aunt Liz’s house and that Aunt Liz and her friends had behaved inappropriately with the girls.
• S.L. considered coming forward with her recantation when she was 19, after she informed a counselor that her testimony had been untruthful, but Javier told her she’d be prosecuted for lying and go to prison, so she did not.
• Before S.L. did come forward with her recantation, Javier told her he would “hit her where it hurts” if she did so. When she did so, Javier complained against her to Child Protective Services regarding her parenting of her own three children.
• Ramirez, Rivera and Mayhugh were clinically assessed (psychosexually evaluated) by the Executive Director of the Counseling Institute of Texas, Maria Molett, who reviews sexual offenders for the State of Texas to determine whether they require civil commitment, and each was determined not to meet the profile of a sexual offender. Anna Vasquez was not seen by Maria Molett, as she had been paroled, but a psychosexual evaluation of her by others employed in a similar capacity was reviewed by her, and that evaluation suggests that she, too, does not meet the criteria.
The above findings of the habeas court are supported by the record.
*298Despite these findings, the habeas court did not recommend granting relief on actual-innocence grounds because, though S.L. recanted, V.L. had not, and her testimony would still be available at trial for the jury to weigh against the recantation evidence. Among the trial court’s conclusions are the statements that
* “There is no hard scientific evidence (such as DNA or the like) which scientifically establishes the guilt or innocence of any of the Applicants.”
* “There are only two eyewitnesses, and one of them continues to assert the truth of her trial testimony.”8
* “The credibility of these two witnesses is an issue for the jury to decide, and no scientific evidence conclusively settles the matter.”
But in this regard, the habeas court appears to have conducted a legal-sufficiency analysis. The legal sufficiency maxims — “a child’s testimony alone is sufficient to support a conviction for sexual assault,” “the jury is the sole judge of credibility and weight to be attached to the testimony of witnesses,” “the fact-finder is free to believe all, part, or none of a witness’s testimony” — echo within these conclusions. However, this is not a direct appeal.
In Elizondo, this Court deliberately stayed away from the issue of legal sufficiency of the evidence in actual-innocence cases because no one would ever be found actually innocent on habeas review if the original trial evidence was legally sufficient to support guilt. As we stated in Elizondo,
On reflection, we now acknowledge that Jackson is not a suitable standard for describing the applicant’s burden of proof in a collateral proceeding where he does not attack the rationality of a fact-finder’s verdict. On the other hand, Justice Blackmun’s formulation, because it focuses on the applicant’s burden of proof, directs the habeas court, as fact-finder, to weigh the newly discovered, exculpatory evidence against the incul-patory evidence offered at trial for the purpose of determining whether it affirmatively shows the applicant to be innocent. Thus, Justice Blackmun continues:
Because placing the burden on the petitioner to prove innocence creates a presumption that the conviction is valid, it is not necessary or appropriate to make further presumptions about the reliability of newly discovered evidence generally. Rather, the court charged with deciding such a claim should make a case-by-ease determination about the reliability of the newly discovered evidence under the circumstances. The court then should weigh the evidence in favor of the prisoner against the evidence of his guilt. Obviously, the stronger the evidence of the prisoner’s guilt, the more persuasive the newly discovered evidence must be.
This is a far more fitting approach to the resolution of factual issues, focusing on the burden and quantum of proof required for an affirmative finding in the first instance rather than on the standard associated with a deferential review of that finding. Accordingly, we now hold that, in the exercise of our postcon-viction habeas jurisdiction under Article 11.07 and 11.071 of the Code of Criminal Procedure, our job is not to review the jury’s verdict but to decide whether the newly discovered evidence would have convinced the jury of applicant’s innocence.
*299Ex parte Elizondo, 947 S.W.2d 202, 207 (Tex. Crim. App. 1996) (quoting Herrera v. Collins, 506 U.S. 390, 443-44, 113 S.Ct. 853, 122 L.Ed.2d 203 (1993)(Blackmun, J., dissenting)). This “weighing” is why a guilty plea does not foreclose an actual-innocence claim. Tuley, 109 S.W.3d at 392 (“The State says that a guilty plea waives any contention regarding the sufficiency of the evidence. This is true, but the State’s assertion that a claim of actual innocence is nothing more than a challenge to the sufficiency of the evidence is not true.”).
The habeas court appears to have concluded that, because V.L.’s testimony would still be available at trial, it would be up to a jury on re-trial to resolve conflicts between V.L.’s testimony and S.L.’s recantation and determine which is more credible. But that is the same analytical framework we rejected in Ex parte Navarijo. 433 S.W.3d at 571-72. There, we denied relief because the habeas court simply performed a one-to-one comparison between the new recantation and the eyewitness testimony without considering the strength of the State’s case as a whole. Id. The same can be said of the habeas court’s analysis in this case.
Indeed, we have previously granted relief on an actual-innocence claim when newly available and credible recantation evidence undermined testimony that would nevertheless have been legally sufficient under the one-eyewitness rule. Thompson, 153 S.W.3d at 420. In Thompson, the defendant was charged with sexual assault of his daughter, and the only direct evidence of the sexual assault at issue that the jury could rely upon was the testimony of the eight-year-old complainant. Id. at 418. Because there was no physical evidence of assault, the State corroborated the complainant’s testimony with the complainant’s mother’s testimony that she became suspicious of the defendant after the complainant came home from time with her father wearing a torn dress. Id. at 419.
But at the habeas hearing, the defendant presented a great deal of evidence involving the on-going custody dispute between the defendant and his ex-wife at the time of the accusations. Id. The complainant herself, a twenty-year-old woman, testified that the sexual abuse never happened, and that her mother had pressured her to testify falsely. Id. The complainant’s mother admitted that she had physically abused her daughter in the past, though she denied there being any custody dispute at the time of the allegations. Id. She admitted that her daughter had originally said that the dress had been torn on a bus, and the church bus driver corroborated this by testifying that she had witnessed the complainant tearing her dress on the bus. Id. Finally, the defendant presented testimony from an expert on interviewing techniques and recantations in child sexual-abuse cases who opined that the complainant’s recantation was valid. Id. at 420. After the consideration of the significant impact that all this evidence would have upon the State’s case, we granted habeas corpus relief because no reasonable juror could have found the defendant guilty.
Though the math may become trickier when, as in this case, there is only one recantation between two accusers, the ultimate calculus should not change. Here, the evidence presented by the Applicants has eroded the persuasiveness of the State’s already weak cases just as surely as the evidence presented at the habeas hearing in Thompson did. In this case, V.L. and S.L.’s interlocking accusations were presented to both juries via Detective Matje-ka, Serafina Limón, Dr. Kellogg and the girls themselves — V.L. at Elizabeth’s trial, and both girls at the joint trial. The accounts were so intertwined that the credibility of one witness’s version rested on the *300credibility of the other witness’s version of events. Both victims testified that they were either assaulted together or one right after the other. S.L.’s habeas testimony that nothing happened cannot be reconciled with V.L.’s trial testimony; a fact-finder could not rationally believe both complainants.9
A. S.L.’s Recantation Is Credible and Corroborated
Notably, S.L.’s recantation is direct, detailed, certain, and strongly corroborated. See Thompson, 153 S.W.3d at 420-21. Her original testimony, like V.L.’s, was not strong to begin with, and her recantation completely undermines her trial testimony. S.L. testified in detail regarding the events that were transpiring at the time that the false accusations came to light, and she explained the underlying motivations that compelled her and her sister to fabricate those allegations. Cf. Franklin, 72 S.W.3d at 678 (actual innocence not shown where child victim recanted testimony that she had never had sex with any other man (other than Franklin); the evidence “calls into question her veracity in general, but only collaterally affects her accusation against applicant”); Cf. Brown, 205 S.W.3d at 547 (actual innocence not proven where child victim claimed a lack of memory and made a global denial of sexual abuse). Rather than provide information that would have merely undermined her credibility, S.L. provided affirmative evidence that the assaults never occurred. In effect, S.L. not only recanted her own trial testimony, she also provided credible eyewitness evidence exonerating the women for the crimes against V.L.
The circumstances and timing of the recantation were not suspicious. As pointed out by Dr. Doyle, the first person S.L. came forward to about her recantation was “a third party who doesn’t have a dog in the show.” It was only when she finally found herself in a comfortable, safe setting that she could deal with her own feelings and thoughts. Cf. Brown, 205 S.W.3d at 548 (“the timing of C.B.’s recantation is, at least on its face, highly suspicious. It was only after applicant’s guilt was adjudicated and he was sentenced to twelve years in prison — more than three years after the alleged offense — that C.B. suddenly told her mother that she had been lying all along”).
The recantation is corroborated by Rosemary Carrillo’s testimony that Javier, with the occasional assistance of his mother, Serafina, had engaged in a pattern of involving the children in false abuse allegations to gain leverage in legal disputes. See Thompson, 153 S.W.3d at 419 (noting that new evidence of an ongoing custody dispute at the time of the accusations undermined the strength of the State’s case); see also Hammer v. State, 296 S.W.3d 555, 569-70 (Tex. Crim. App, 2009) (noting that a history of making prior false allegations against others could be probative evidence of whether allegations at issue in the case are false under the doctrine of chances). Moreover, S.L.’s testimony, credited by the trial court, establishes that Javier repeatedly threatened S.L. with prison when she first tried to come forward with her recantation. This corroborated S.L.’s ac*301count that her father had used threats to manipulate her into making the sexual assault allegations in the first place, and suggested a reason why V.L. had not recanted her testimony.
Additionally, S.L.’s recantation is consistent with the new medical testimony establishing that there was no physical evidence of sexual assault. Thompson, 153 S.W.3d at 420 (noting that recantation testimony was consistent with the lack of physical evidence of abuse from the physical examination); Cf. Navarijo, 433 S.W.3d at 571 (clear and convincing standard not met in light of the still-standing medical evidence suggestive of sexual abuse that was presented at trial). The combined force of S.L.’s credible testimony and the new scientific evidence that there are no physical signs of sexual assault distinguishes this case from the situation presented in Na-varijo where the credible recantation testimony remained contradicted by viable medical evidence that a sexual assault had occurred. Navarijo, 433 S.W.3d at 570-71. And, as we have already held, the Applicants would not have been convicted of these offenses had they been allowed to consider the newly available, relevant scientific evidence regarding the victim’s injuries.
Finally, Applicants presented expert psychological evidence to further explain why S.L.’s recantation is both credible and very persuasive. In deciding whether to grant habeas claims of actual innocence, this Court has considered the testimony of experts who have training in the detection of false sexual-abuse allegations and false recantations. C.f. Ex parte Harleston, 431 S.W.3d 67, 79, 88 n. 11 (Tex. Crim. App. 2014) (denying relief on an actual-innocence claim, in part, because the testimony by K.D., the recanting witness at the live habeas hearing, was “internally inconsistent and present[ed] implausible explanations” of why K.D. would have fabricated sexual-assault allegations against Harle-ston, and in part based on the State’s evidence at trial that included “a number of witnesses who supported the circumstances of KD.’s sexual-assault outcry as . genuine.”). Here, the Applicants’ psychological experts detailed their reasons for finding S.L.’s recantation significantly more credible than the trial testimony that was introduced in this case.
Specifically, Dr. Doyle, testified as to her belief that S.L.’s recantation testimony was credible both in terms of the substance of the recantation and in the manner in which it came to light. She explained that psychological studies have shown that telling a child a false story repeatedly eventually taints the child’s memory. Dr. Doyle also explained that this could have been exacerbated by the investigation itself. At the time of S.L.’s interview by police, there was little research into effective interviewing techniques for children who claim to have been sexually abused. Unlike more contemporary investigations into child abuse allegations, in which a trained forensic interviewer asks a child open-ended questions in a one-on-one interview, Dr. Doyle noted that S.L. had sometimes been interviewed in front of her father and alongside her sister. These techniques, according to Dr. Doyle, are associated with inducing erroneous responses and increased the likelihood of a child acquiescing to misinformation provided by the interviewer or the parent. Dr. Doyle’s expert testimony further supports the determination that S.L.’s recantation is both persuasive and credible while simultaneously bringing V.L.’s trial testimony further into question.
Additionally, Applicants presented unchallenged testimony from the current executive director of the Counseling Institute of Texas, Maria Molett, that none of the *302Applicants had ever engaged in any behavior similar to the allegations in this case, further corroborating S.L.’s testimony that the assaults never occurred. Molett testified that she had previously spent ten years as a member of the governor-appointed Council on Sex Offender Treatment, which is the organization that makes determinations and policies regarding civil commitment of sexually violent predators. She explained that she had determined who could become a licensed sex-offender-treatment provider for the State-sponsored sex-offender-treatment programs in her capacity as chair of the education committee as well as a member of the Standards and Practice committee. She also noted that she is a licensed sex-offender-treatment provider.
Molett recounted that she had conducted psychosexual evaluations on three of the Applicants, Liz, Cassie, and Kristie, all of whom were still in prison when these allegations were being investigated. As part of this evaluation, she reviewed all of the records of the offenses for which the Applicants were convicted. She did a risk assessment, a Hare Psychopathy, and community supervision risk assessment level of service inventory. She also testified that she utilized polygraph tests as a means of corroborating the information given to her by each Applicant during these evaluations. According to Molett, each Applicant she tested had passed the polygraph tests. She also explained that though she did not perform an evaluation of Anna, she examined the records and the test results from the person who had evaluated Anna. Mol-ett concluded that based upon these evaluations, none of the four women were sex offenders and none of them had engaged in deviant sexual behavior. As mentioned above, the habeas court expressly found Dr. Mollett’s opinion evidence credible.
Applicants urge this Court to not only credit Molett’s expert opinion testimony relied upon by the habeas court, but also the exculpatory results of the polygraph tests themselves as proof of their innocence. We have previously held that polygraph results themselves are inadmissible because they are unreliable. Leonard v. State, 385 S.W.3d 570, 577 (Tex. Crim. App. 2012) (“For more than sixty years, we have not once wavered from the proposition that the results of polygraph examinations are inadmissible over proper objection because the tests are unreliable.”). In the context of an actual-innocence claim, we have refused to consider them as evidence of innocence. Miles, 359 S.W.3d at 662 n. 14 (“Because polygraph exams are not admissible evidence, we do not rely on these results as evidence of Applicant’s innocence.”). And while we have acknowledged that Rule 703 allows an expert to base his or her opinion on otherwise inadmissible evidence, we specifically declined to allow this rule to be used to admit polygraph results by claiming that the results themselves were necessary to support the expert’s opinion. Leonard, 385 S.W.3d at 582. As we explained in Leonard, Rule 703 cannot be used a conduit for admitting opinions based on “scientific, technical, or other specialized knowledge” that would not meet Rule 702’s reliability requirement. Id. Though we seemed to leave open the possibility that an expert could base his or her opinion on a polygraph result, we refused to admit the results of a polygraph test under the guise of an expert’s opinion that was based solely upon the results themselves.
We do not need to rely upon the polygraph results themselves in these cases to determine whether the Applicants unquestionably established their innocence. First, it would require this Court to overrule existing law. Applicants are trying to do exactly what we said the State could not *303do in Leonard, use polygraph test results to prove the truth of those results under the guise of expert opinion testimony based on those results. Applicants have not shown that our previous precedent was wrongly decided or unworkable.
Second, the habeas corpus court clearly credited Molett’s opinion that the Applicants in this case were not sex offenders, that they had never engaged in deviant sexual behavior, and that they were not at any risk to engage in any such behavior in the future. Molett’s opinion was based upon a whole series of evaluations. The polygraph results were used only to corroborate the information provided by the Applicants in some of those tests as part of an overall treatment evaluation. 22 Tex. Admin. Code § 810.64(c)(18) (“polygraph examinations shall be used as a part of a comprehensive treatment program”). Even if the results themselves were credited, they would only provide, at most, incremental proof of the Applicants’ claims beyond what the habeas court has already credited. Given how strong the new evidence of innocence is when compared to how weak the remaining evidence of guilt is, Applicants’ claims for relief simply do not turn upon the polygraph results themselves. If the results of lie-detector tests were truly the lynchpin of the Applicants’ claims, it would be difficult to argue that Applicants have carried their Herculean burden to prove their claim of actual innocence. With a number of convictions in Texas being overturned due to the fact-finder’s reliance upon “junk science,” we see no reason to open the door to consideration of historically unreliable polygraph results themselves just because those results might either enhance or discredit exoneration claims.
In these cases, it is unnecessary to do so. Applicants have presented significant new evidence that unquestionably establishes their innocence. S.L. not only established that the offenses did not occur through her credible recantation testimony, she explained in detail how her father forced her and her sister to make the false allegations to the police in the first place. S.L.’s recantation is corroborated by other documented instances of S.L.’s father fabricating allegations of abuse in order to manipulate his wife in an ongoing custody dispute. Dr. Doyle’s expert testimony regarding false claims of abuse further corroborated S.L.’s recantation, and Maria Molett’s expert opinion supports the Applicants’ claims that they are not sex offenders and the alleged behavior is completely inconsistent with their psychosexual histories and psychological evaluations. All of this evidence combines to paint a clear and consistent picture of the Applicants’ innocence.
B. The Remaining Evidence of Guilt is Exceedingly Weak
In contrast, the evidence relied upon to convict the Applicants paints a fairly inconsistent picture establishing the Applicants’ guilt. The evidence presented at the two trials set forth multiple different versions of how the Applicants were alleged to have abused V.L. and S.L. during the time they visited their Aunt Liz’s apartment. In one version both complainants were together in the same room during the assaults, but in another the complainants were not in the apartment together because S.L. was locked outside while the Applicants purportedly assaulted V.L. inside the apartment. At one point, S.L. claimed she saw V.L. leave the bedroom to take a shower after allegedly being assaulted. In another, V.L. left the bedroom to simply go outside and play immediately after the alleged sexual assault. Sometimes it was a gun to V.L.’s head when she was on the phone to her father, others it was a knife. The stories also vary as to the claims about who *304held the complainants down and who actually committed the sexual assaults. None of the versions are consistent regarding when these two events allegedly occurred during the week in question. These material conflicts are so great that it is difficult to tell which version of events the jury believed.
Most importantly, many of the details of these stories are simply implausible. As Dr. Doyle, explained in the habeas hearing, these stories simply did not make sense. Dr. Doyle could not find one piece of evidence or one statement consistent with what one would expect in a true sexual-abuse allegation. According to Dr. Doyle, the events described were more like a fraternity hazing without any of the elements of sexual abuse. For example, Dr. Doyle explained that unlike a typical family sexual-abuse scenario there was no evidence in this case of grooming or an attempt to isolate the children. Having interviewed hundreds of people who had been sexually abused, Dr. Doyle testified that she had never seen anything that came close to the alleged behavior in this case. That is why she believed that this story was a generated story; it did not hang together. S.L.’s testimony did. The State did not challenge this testimony.
Given the implausible and contradictory nature of the allegations, Dr. Kellogg’s medical testimony was crucial to the case. It was the only piece of evidence that could show a crime had ever occurred. Not surprisingly, this opinion testimony was central to the State’s theory of criminal liability. From opening argument, to closing argument, there were two themes in both trials: the girls’ stories were inconsistent, but that didn’t matter in light of Dr. Kellogg’s testimony.
At Elizabeth’s trial, the prosecutor dealt with the coming inconsistencies head-on stating, in an opening argument (that was made before it was known that S.L. would not testify), that “they’re not going to be able to remember everything. And they may not be able to keep everything perfectly straight, but they’re going to do their best to tell you what happened.” But “you will hear from Dr. Nancy Kellogg. .,. who has done it, if not thousands, at least hundreds and hundreds of rape examinations on children... And she will talk to you about ... the medical finding which [was] consistent with the story that V.L. told about how Liz — Aunt Liz and her friends raped her.”
V.L. went on to testify in that trial that, immediately after the first assault, her father called and she answered the phone. Liz then pulled out a gun and pointed it to her head and said that “if I told anybody that she was going to kill me and my family.” She was sure “100% sure” it was Elizabeth.
Dr. Kellogg testified that, in taking the history from the girls, V.L. told her that both Liz and Anna had guns. Serafina, the outcry witness, testified that V.L. told her that “one of the girls had the knife and had told them — threatened them that they were going to do something to their dad and me.” And in her statement to police, V.L. said that it was Cassandra who made the threat to kill.
V.L. was cross-examined about these inconsistencies, as well as ones relating to whether the two assaults happened on sequential or non-sequential days, whether or not S.L. was present during the outcry to Serafina, whether Elizabeth drank nothing or half a bottle of Tequila, whether V.L. herself was made to drink wine with soda or nothing at all, whether S.L, was in the living room playing during the first assault or whether she was locked outside, whether they were made to go back outside to play after those assaults or whether they stayed in the apartment, whether the *305four women were there the entire week or not, whether the women were screaming at them or not, whether liquids were inserted both days or just the first day, whether it was Anna who took off her pants or Kristie, whether the gun was held to her head or to both of their heads, whether S.L. was assaulted in the living room or the bedroom, whether the first assault was in the morning or at 2 p.m., and whether she saw tattoos and on who.
The State dismissed the inconsistencies in light of Dr. Kellogg’s testimony.
You heard lastly from Dr. Kellogg. I think Dr. Kellogg, ladies and gentlemen, interestingly enough was perhaps one of the most critical, if not the most critical, witness that you heard from.... It’s interesting to me that after all of the emotion and after all the technical wizardry, it’s this simple and this terrifying painful. That little tag, that scar at 3:00 o’clock, don’t ever, ever forget Dr. Kellogg’s words that scar, that healed tear is indicative not just of penetration. It is indicative of very painful penetration. Folks, we can sit here all day long and talk about demeanor and we can talk about well, the inconsistencies they had were this. The inconsistencies they had were that. But the medical, physical evidence does not lie. You can’t make that tag — you can’t make that painful tear, that painful healing, that painful scar, up. That’s why we brought you Dr. Kellogg.
A year later, at the joint trial in 1998, V.L. testified that it was Anna who had put the gun to her head and her Aunt Liz who had answered the phone after the alleged assaults. She denied that she had said, in the 1997 trial, that it was Liz who had the gun.10 V.L. was again thoroughly cross-examined on the variations of the facts in her accounts, including whether the Applicants were dressed or partially undressed or fully undressed, whether she made an outcry to her grandmother because of pressure she felt or she did so in response to her grandmother’s questioning about how the girls were playing with dolls, whether the door to the outside was locked with a chain lock or regular lock, whether she talked to her father about the assaults or whether her grandmother told their father, whether or not Kristie was there the first day, whether or not Kristie was there the second day, and whether they were taken home or picked up. V.L.’s inconsistencies about significant details of the offense such as whether the defendants used a gun or a knife, who was or was not present, and where the assault occurred are represented in this chart.
*306[[Image here]]
Although the court of appeals dismissed all the inconsistencies because none were on the “essential elements,” it did so because the essential elements were corroborated by Nancy Kellogg’s testimony regarding the physical evidence of sexual assault. See Mayhugh v. State, 1999 WL 1246925 at *4 (Tex. App.-San Antonio Dec. 22, 1999) (not designated for publication).
*307Dr. Kellogg has now recanted that testimony as scientifically unreliable. S.L. has recanted her testimony as well, admitting that the allegations were fabricated. Indeed, S.L. explained exactly why the girls were able to remain consistent on those points and no others: their father would hit them if they did not recount the story correctly. And finally, it is logically impossible to believe any one of V.L.’s versions if one believes S.L.’s recantation. The State’s theory of the case was that the girls were witnesses to each other’s alleged abuse. If S.L. was not abused, neither was V.L. Without Dr. Kellogg’s or S.L.’s testimony, the State is left with an exceedingly weak case from one witness that must stand against significant evidence of witness manipulation and eyewitness testimony that the crimes never occurred.
Applicants have presented considerable and extremely persuasive evidence to support their claim of innocence. The medical testimony relied upon to secure the convictions is now known to be unreliable. One of the complainants has not only recanted her own testimony, she has provided eyewitness testimony that no assaults ever occurred and that she and her sister were forced to testify falsely against the Applicants. New psychological evidence corroborates Applicants’ claims that these allegations were generated through the manipulation of the complainants’ by their father in order to gain leverage in a custody dispute. Substantial evidence regarding a history of claims of abuse brought forth by the complainant’s father that we now know were equally false further corroborates the recantation evidence in this case. And evaluations of each Applicant showing that they are not sex offenders and have never engaged in deviant sexual behavior, further establishes the claims of actual innocence. When this new evidence is compared to the State’s exceedingly weak case for guilt, it is patent that the Applicants have unquestionably established their claim that no jury could rationally find them guilty.
VI. Conclusion
It has been suggested that the term “actual innocence” is inappropriate because applicants who are successful when raising a claim of actual innocence never truly prove that they did not commit the offense. But when the presumptions are reversed, the State does not have to prove that a defendant is definitively guilty. The State does not prove that a person has committed a crime beyond all doubt, or even beyond a shadow of a doubt. By proving its case at trial according to the applicable standard, the State secures the ability to proclaim to the citizens of Texas that the person responsible for a crime has been brought to justice, that the person is guilty. When defendants have accomplished the Herculean task of satisfying their burden on a claim of actual innocence, the converse is equally true. Those defendants have won the right to proclaim to the citizens of Texas that they did not commit a crime. That they are innocent. That they deserve to be exonerated. These women have carried that burden. They are innocent. And they are exonerated. This Court grants them the relief they seek.
Keller, P.J., and Keasler, J., joined majority opinion as to part three. Alcala, J., filed a concurring opinion in which Meyers, J., joined. Hervey and Yeary, JJ., did not participate.. The testimony in both trials is not very specific with regard to when the events were have alleged to have taken place. At least one version of events places the alleged assaults in August of 1994.
, Ms. Limon’s first name is spelled two different ways: Serifina and Serafina. We use Ser-afina for consistency’s sake. Also Elizabeth Ramirez is sometimes referred to as Liz and Cassandra Rivera is sometimes referred to as Cassie.
. Generally, expert testimony that child victims of sexual abuse have provided truthful *291testimony is inadmissible. Yount v. State, 872 S.W.2d 706, 712 (Tex. Crim. App. 1993). We have held that testimony that a child-abuse victim did not exhibit “behaviors that point to being manipulated” was not a direct comment on the truthfulness of the child victim’s allegations. Schutz v. State, 957 S.W.2d 52, 73 (Tex. Crim. App. 1997). But here, Dr. Kellogg’s opinion testimony was specific to the child's statements and characterized V.L.’s testimony as truthful. Thus, Dr. Kellogg’s characterization of V.L.’s allegations as “guileless” and “uncontrived” would run afoul of the general prohibition against admitting such testimony that we announced in Yount.
. Though it was never revealed to the jury in either trial, the 1992 sexual assault exams were the result of an allegation of sexual abuse orchestrated by Javier Limón in Colorado prior to the allegations of sexual assault in these cases. These 1992 allegations concerned claims that V.L. and S.L. and been sexually assaulted by a ten-year-old babysitter. At the hearing on the Applicants’ writ applications, these allegations were revealed to be false and that no such child had ever existed.
. Although Dr. Kellogg did not expressly recant her testimony about S.L.'s physical exam, the parties and courts appear to agree that her testimony about the "redness” was non-consequential. As noted by Applicants, Dr. Astrid Heger, a nationally respected and recognized expert and colleague of Dr. Kellogg's, reviewed the evidence, testimony, and photographs and concluded that she could say “to a medical certainty, that neither V.L. nor S.L. showed any physical indicators of past physical trauma to their hymens or any other part of their sexual organs. Specifically, the September 28, 1994 photographs showed that both S.L. and V.L., as of that date, had perfectly normal hymens, with no physical indications of past traumas whatsoever.”
. The late Judge Mike Machado had presided over Elizabeth Ramirez’s individual trial.
. Though Ms. Mollett opinions were based in part on polygraphs, she testified that her opinions were also based upon research-based risk-assessment tools used in the practice of assessment and treatment of sex offenders.
. Applicants note that this conclusion is based, solely, on the fact that, when V.L. was 15, she told S.L. that the abuse “may not have happened to you but it happened to me.”
. It is worth noting that the habeas court has already reached a similar conclusion with regard to the impact of Dr. Kellogg's ''recantation.” As discussed above, the habeas court found that the testimony of both complainants was so inextricably intertwined that Dr. Kellogg's testimony corroborated the evidence supporting the crimes against both girls. Without that corroboration, the habeas court held, that the State’s case was weakened enough that it was "more likely than not” that a rational jury would have acquitted the Applicants.
. Serafina testified at the 1998 trial that the girls had told her about a gun, but that she had forgotten to tell the police about the gun. She said she never mentioned a knife in her 1997 testimony. She, like V.L., blamed the court reporter for the discrepancy. S.L. told prosecutors in 1996 that Kristie held a gun on them.