Update: On June 16, the Texas Court of Criminal Appeals stayed Roberson's execution, sending his case back to a trial court for a hearing on the new scientific evidence.
The scientific evidence was conclusive, doctors told a Texas jury in 2003: capital murder defendant Robert Roberson had violently shaken his two-and-a-half-year-old daughter to death.
Thirteen years later—and just days before Roberson is scheduled to be executed—four medical experts are now claiming that the scientific theory used to convict him has been thoroughly debunked.
Roberson, 49, is the next death row inmate in America scheduled to be executed, and will go to the death chamber June 21. He was sentenced to death in 2003 for the murder of his daughter, Nikki Curtis.
Last week, he filed an application for a writ of habeas corpus—essentially an appeal to stay his execution and be given a new hearing—to the Texas Court of Criminal Appeals, the highest criminal court in the state. A decision is expected within days.
At the time of Roberson's trial, doctors believed that certain symptoms in a child could conclusively prove they were violently shaken or abused, based on a theory known as Shaken Baby Syndrome. But in the last decade, the four experts who submitted affidavits as part of his appeal say, there's been a sea change in the scientific understanding of the issue.
Now his best last hope to postpone an execution and get a new hearing is Texas' junk science law, which guarantees defendants a new trial if they can prove they were convicted based on scientific evidence that has since been discredited. It's seen as one of the most progressive laws in the country of its kind.
Roberson, whose lawyers say suffers from "severe limitations in intellectual functioning," was Nikki's biological father but essentially shared custody with the parents of Nikki's mother, whom she had lived with for the first two years of her life. He brought her home from her grandparents house in Palestine, Texas, on the evening of January 30, 2002, and put her to bed. Early the next morning, he was woken up by her crying, and found that she had fallen on the floor. She seemed OK, he told investigators, so he put her back in her bed and went to sleep. When he woke up again a few hours later, she was blue and barely breathing. Roberson rushed her to the hospital, and she was declared dead the next day.
Prosecutors dismissed Roberson's account, his lawyers say, and instead charged him with murdering his daughter by shaking or beating her. At the time, most doctors believed that they could determine that a child could be diagnosed with Shaken Baby Syndrome based on three symptoms: retinal hemorrhaging, subdural hematoma/hemorrhaging, and edema, or brain swelling. Roberson's jury was told that because Nikki had signs of all three, she must have been abused.
The defense didn't contest that explanation or call any medical experts. On February 14, 2003, the jury convicted Roberson and sentenced him to death.
Now, however, a growing group of scientists disagree with this method of diagnosing Shaken Baby Syndrome. Research has shown that the same symptoms can be caused by other natural or accidental causes, the four experts who reviewed Roberson's case write. They present a range of possible causes for Nikki's death that were not explored during the trial: meningitis due to an ear infection; an injury before Roberson arrived; a short fall like the one he described or a congenital condition.
"The medical personnel in this case seemed to adopt the presumption that child abuse was the root cause from the outset, a presumption that appears to have imposed a cognitive bias that prevented them from investigating all reasonable possible causes of death either before or after her death," wrote Dr. Harry Bonnell, a forensic pathologist.
Notably, the experts—who also include two other doctors and a professor of mechanical engineering—agreed that "it is impossible to shake a toddler to death without causing serious neck injuries—and Nikki had none," the appeal states.
Across the country, there are several hundred people in prison for murder convictions based on Shaken Baby Syndrome, L.A. Weekly reported, even though the latest scientific findings would undermine their cases. The American Academy of Pediatrics now doesn't even use the term "shaken baby syndrome," based on an understanding that its effects are not only caused by shaking. (A smaller contingent of doctors, however, still believe the kind of evidence used to convict Roberson is conclusive.)
The prosecution in Roberson's case also originally charged him with sexually assaulting Nikki, and several witnesses talked about that theory in front of the jury. But no evidence was presented that any sexual assault took place, and the prosecution dropped the charge at the last minute. Roberson's lawyers argue that was highly prejudicial. "The State used this rank speculation to drive home its view that Robert was not just a poor, mentally impaired father struggling with sobriety, but a deviant—capable of raping and brutally shaking his own daughter to death," they write.
Gretchen Swenn, Roberson's attorney, declined to comment while the case is being considered by the court.
Last year, Roberson asked the U.S. Supreme Court to remove his previous lawyers, James Volberding and Seth Kretzer, alleging that they had failed to adequately represent him in his appeals. The justices declined to hear his case.
That means that this appeal is most likely Roberson's last chance at a stay of execution. He has an application for clemency pending at the Texas Board of Pardons and Paroles, but clemency for Texas death row inmates is exceedingly rare.
The state has not yet filed a response to Roberson's appeal. The Court of Criminal Appeals is expected to rule on his case in the next few days.
"When the trial record is viewed through the lens of current science and evidence-based medicine, it is clear that he is innocent of capital murder," Roberson's appeal states.
Casey Tolan is a National News Reporter for Fusion based in New York City.