Floresbinda Plata hadn’t seen a doctor during her entire pregnancy in the desolate village of Angoa in Michoacan, Mexico. But after four hours of painful labor, she sought help at the nearest clinic, an hour away by dirt road. After Plata arrived, Dr. Luis Zapien recalls, “We pulled [the baby] out and he was born completely flaccid and purple.”

Floresbinda heard the doctor say that her son was dead before he untwined the umbilical cord that was wrapped twice around the baby’s neck and began mouth-to-mouth resuscitation. After several minutes, though, her son began breathing. But the lack of oxygen had already damaged his brain. A nurse checked off a simple behavioral checklist — did he cry, did he respond appropriately? — and gave him two points out of 10, a score for a newborn with profound cognitive defects. Just an hour into his life, and 20 years before he would be sentenced to die in Texas, Daniel Plata was already being tested for mental retardation.

By the time he was 3, Daniel could say “Mama” and “Papa,” but not much else. His grandmother grew frustrated when, as he got a little older, he couldn’t seem to run simple errands. “If I sent him for lard he would lose the money,” says Cynthia Hernandez. “If I sent him for peppers he would bring back tomatoes.” In school, Daniel stood out as a slow learner. His first-grade teacher, Eleazar Herrera Solis, “tried to get him to be the same as the rest,” but “the child could barely read.” His violent father complicated matters. Several times a week he would come home drunk and attack Floresbinda. As the oldest child, Daniel would try to protect his mother and two brothers from Isidro’s fists, belt and occasionally his machete. In the process he became the target of his father’s rage.

In 1986, Floresbinda fled with her sons to the United States, hoping for safety and a better life. She found work as a janitor in Houston. When the boys registered for school, Daniel — then 10 — was put in first grade. His friend Nasario Vasquez remembers him as “the kid who got picked last” for basketball. “For Daniel the games had no rules,” Vasquez says. “He would just run down the court and throw up a crazy shot with no coordination.”

When Daniel was 15, he was socially promoted to the ninth grade. He acted up in class and was sent to an alternative learning center. He was flagged as “extremely low” performing by his teacher, Terry Rizzo, in a note to the school counselor. At first Rizzo assumed Daniel was having trouble understanding English, but after studying his behavior, she thought he might be learning disabled. She urged the school to test Daniel to see if he should be placed in special classes. But he was never tested and before his ninth-grade year was halfway over, he dropped out.

Daniel started working as a busboy at Luby’s to help support the family. He took to carrying his mother’s gun around as a way to look tough. Then one night in March 1995, Daniel brought the gun along when he and some friends went to rob a nearby Stop’n Go.

He had drunk about 20 beers and smoked PCP-laced marijuana, he later testified, so his memory of the night is hazy. But the store’s security camera shows Daniel pointing his gun at the clerk, Murlidhar Mahbubani, and yelling, “Give me the money!” His two friends jumped over the counter and emptied the cash register of about $50. Then Daniel bent over the counter and shot Mahbubani several times in the back.

The store’s surveillance system clearly videotaped his face. It also showed him, on the way out, using his shirt to wipe his fingerprints off the door.

Within 30 hours, police had Daniel in custody. He confessed to Mahbubani’s murder soon afterward.

During his trial in 1996, prosecutors repeatedly played the videotape showing Daniel ruthlessly killing Mahbubani. The guilty verdict was a foregone conclusion. During the penalty phase of the trial, Daniel’s mother and stepfather testified that he was a good son, and his attorney argued that he was “passive, docile. For one minute and a half, he just lost it.” In the prosecutor’s closing statement, he urged a death sentence: “This was a shocking crime, and it deserves a shocking punishment.” The jury agreed. Daniel Plata was sentenced to die by lethal injection.

Within four years, Plata’s appeal had wound its way through the courts and ended in failure. His mother sought help from several lawyers in Mexico; her inability to speak English made it hard for her to find legal assistance in the United States. “I would sleep and wake up with the same thought about each day passing… that he was one day closer to death,” she says.

Two more years passed before officials from the Mexican Consulate in Houston called her and said a lawyer wanted to ask about Daniel’s history of being slow. The lawyer thought it might save his life.

In 2002, six years after Daniel Plata landed on Death Row, the U.S. Supreme Court ruled in a case called Atkins v. Virginia that “executions of mentally retarded criminals are cruel and unusual.” Even though mentally disabled people can understand the difference between right and wrong, the court reasoned that they are less able to control impulsive behavior or learn from mistakes. The court supported its decision by pointing to bans on executing the mentally retarded in 17 states and in federal cases as “evolving standards of decency.”

Like most of the states that had already passed bans, the justices used a clinical definition to establish the level of mental retardation that would exempt Daryl Atkins, the Virginia defendant, from death: below-average intellectual abilities defined by an IQ score of 70 or below and “deficits in adaptive behavior” such as practical and social skills. Both of these limitations, the court ruled, had to be present before the age of 18.

But the court left it up to the states to choose their own definitions of mental retardation. Since 2002, eight more states have passed laws that use the clinical definition cited in Atkins. Texas is not one of them. With bipartisan support, the Texas Legislature passed a law in 2001 mandating a life sentence for mentally retarded people convicted of capital crimes. But Gov. Rick Perry vetoed the measure, agreeing with critics that it was a “backdoor attempt to ban the death penalty.” Bans on executing the mentally retarded have been floated in every legislative session since but have never again come up for a vote.

In 2004, a Texas death row inmate named Jose Briseño contended that he was mentally retarded and shouldn’t be executed for murdering a Dimmit County sheriff. In the absence of legislative guidelines, the Texas Court of Criminal Appeals wrote “temporary judicial guidelines” that have guided Texas courts ever since. In its Briseño decision, the court called clinical definitions of mental retardation, like those used by the U.S. Supreme Court, “exceedingly subjective.” Texas added its own set of additional criteria in the form of seven questions, including: “Did the commission of that offense require forethought, planning and complex execution of purpose?” If a defendant didn’t address these questions to the court’s satisfaction, he could be eligible for execution even if his test scores showed he was mentally disabled.

Most of Texas’ questions emphasize the events of a crime in deciding whether a defendant meets a legal definition of mental retardation. “I think much of that emphasis is inappropriate because it embodies the stereotype of mentally retarded people as unable to do anything,” says Sheri Lynn Johnson, a professor at Cornell Law School and co-director of its Death Penalty Project. In Texas, under the Briseño standard, if you’re capable of committing a murder, it’s difficult to establish that you’re also mentally retarded.

In other states, evidence of mental retardation is heard in pretrial hearings that decide whether a person is even eligible for a death sentence. In Texas, prosecutors have fought successfully to hold off evidence of mental retardation to the penalty phase of a trial, meaning that jurors consider it only after they have convicted a defendant of murder. Keith Hampton, legislative director of the Texas Criminal Defense Lawyers Association, says “the gamesmanship is this: I can make you hate this guy so much that you won’t care if he’s mentally retarded.”

Since 2002, Texas has removed just 13 men from Death Row after they were found to have the mental and emotional development of 12-year-olds. In contrast to a 40 percent success rate for Atkins appeals nationally, just 28 percent have been successful in Texas. “I suppose you could imagine that Texas Death Row inmates are smarter than everyone else,” says Johnson, “but I’d be surprised.”

During Daniel Plata’s original trial, prosecutors had portrayed him as a sophisticated criminal who’d tried to hide his identity and erase his fingerprints after murdering Murlidhar Mahbubani. But attorney Kathryn Kase figured that Plata’s accomplices, who made no such attempts, had realized the store’s security camera had captured their faces and didn’t bother. If anything, she thought the crime showed how Plata was prone to act impulsively, as mentally retarded people are known to do. And when she interviewed Floresbinda Plata, she learned that there was a family history of retardation: Daniel’s younger brother, Jesus, and his Aunt Celianel had both been diagnosed as mentally retarded. His cousin, Rosalba, had Down syndrome.

To prove that Daniel Plata should be exempt from the death penalty, Kase had to start by showing that he had an IQ of 70 or below. She hired Antonin Llorente, a neuropsychologist who had designed intelligence tests and was a native Spanish speaker — important because it would allow him to test Plata in the language he understood best.

In May 2003, Llorente spent about five hours with Plata in a small visiting room at the Polunsky Unit in Livingston, where men on Death Row are housed. He began by asking Plata if he felt he was mentally retarded. Plata vehemently denied it. When Llorente asked him to draw his family, the 28-year-old man “drew stick figures,” which Llorente noted in his report were “appropriate for children, not mature adults.” Then he measured Plata’s intellectual ability through puzzles and math questions that are part of a test called the Wechsler Adult Intelligence Scale.

Llorente reported that Plata’s IQ score was 65. Even in Texas courts, it’s generally accepted that IQ scores include a “standard error of measurement” of five points up or down. This means a person’s IQ score falls within a range; a person who tests at 75 could still be considered retarded. Plata’s 65 was a strong indication that his intellectual abilities were below average and met the U.S. Supreme Court’s standard for mental retardation.

The next psychologist to evaluate Plata was Texas prosecutors’ favorite tester, George Denkowski of Fort Worth. Denkowski’s career stretched back 30 years to when he directed a 15-bed group home for mildly retarded adolescent offenders in Houston, teaching them adaptive skills that would improve their behavior. He’d also been the chief psychologist at the Fort Worth State School, a 365-bed facility for people with all ranges of mental retardation. Since 1989 he’d been in private practice conducting psychological evaluations. Denkowski had also directed a national study of mentally retarded people in state prisons.

After the Atkins decision in 2002, Denkowski became the first choice for Texas prosecutors. He would ultimately testify in 29 cases — nearly two-thirds of such appeals in Texas to date. In one of the first cases he worked on, Denkowski found James Clark, a man accused of raping and killing two teenagers in Denton, mentally retarded. The state dismissed him after that finding and hired another expert who disagreed. Denkowski’s opinion was presented by the defense to no avail, and Clark was executed.

In 29 cases, Denkowski has found defendants retarded only eight times. By 2006, when he tested Plata, Denkowski had garnered an “almost Dr. Death status” among defense lawyers, according to attorney Robert Morrow. Morrow represented Alfred DeWayne Brown during his 2004 trial for killing a clerk and a security guard at a Houston check-cashing store. Morrow said “Denkowski pretty much thought that if you had engaged in criminal behavior you were not retarded,” Morrow says. Brown remains on Death Row.

The work was lucrative. Denkowski charged prosecutors hourly rates of $180 for evaluations, and $250 for court testimony. Most of the cases he worked on were in Harris County, which until 2009 pursued more death-penalty sentences than any other county in Texas. Between 2003 and 2009, Harris County paid him $303,084 for his services, according to the Harris County Auditor.

Denkowski did not respond to repeated interview requests for this story.

When “Dr. Death” met Daniel Plata at the Harris County jail over the course of two days in March 2006, each of the roughly five-hour sessions was videotaped at defense attorney Kathryn Kase’s request. Like the video of his behavior in the Stop’n Go robbery, this footage would later play a crucial role in deciding whether Plata would be executed.

A large desk dominated the small, brightly lit room where the two met. On one side sat Denkowski, a slim, middle-aged, mustached man. On the other side of the desk, Plata, wearing his orange prison jumpsuit with “JAIL” stamped on the back, sat up straight in his plastic chair and paid close attention to Denkowski, who spoke slowly and maintained a patient, yet authoritative, demeanor. On a small table next to the psychologist sat several open binders and test booklets.

Denkowski, who does not speak Spanish, began by introducing himself and handing Plata his business card. But when he asked Plata to read the court order that had led to their meeting, things got off to a slow start. Plata stared at the page and asked for clarification, saying, “I don’t understand this word right here.” Denkowski explained: “‘Evaluate,’ that’s like, do testing. Like your attorney said. That’s another word for testing.”

Plata: “Do I have to read it all?”

Denkowski: “No. Just, do you… is it OK with you? Do you understand what it means?”

Plata: “Yeah, um.”

Denkowski: “What do you think it means? Tell me.”

Plata: “That you’re uh. Represent. That you’re uh, hire by the state. Something like that.”

Denkowski then ran through several rounds of memory questions designed to trip up people who are intentionally trying to do poorly. Plata scored well, convincing Denkowski that he was giving his best effort. And since Plata seemed to understand him well enough, Denkowski assumed it was fine to do the IQ tests in English. But when Plata misread the word “jar” as “job” on a reading test, Denkowski figured he’d just misspoken and noted his “Hispanic mispronunciation of ‘j.'” The extra point Plata received for this answer bumped up his score from a fifth-grade reading level to a seventh-grade level. It was the first indication that Denkowski was making assumptions during the testing that would inflate Plata’s scores.

As the evaluation continued, Denkowski began to reword questions from the tests as he read them to Plata — a no-no in administering these tests. One of the questions designed to measure comprehension is, “What are some reasons a defendant would choose to be tried by a jury of peers?” But Denkowski, apparently realizing that Plata had trouble with more complicated terms, didn’t use the word “defendant.” Instead, he asked, “Why is it important to be tried by a jury of your peers?” Plata replied: “Because the jury don’t know nothing about the law? I mean that’s exactly what it is, right?” Denkowski asked the question again, differently: “OK, what’s another reason a person should have a trial by jury?”

According to the manual for administering these tests, Denkowski’s rephrasing of the question was a violation of procedure. The irregularities didn’t end there. The video shows Denkowski assigning more points than he should have for some of Plata’s answers. During a vocabulary test, he asked what the word “winter” meant. Plata’s reply — “When it’s cold… when it snows… when we wear jackets” — should have been worth one point, according to the manual. A two-point answer would have been either “the cold season (of the year),” or “the season of the year between fall and spring.” Denkowski gave Plata two points anyway.

After straying from proper administration and scoring methods, Denkowski inflated Plata’s scores through “estimation.” Denkowski used his assumptions about Plata’s upbringing to re-score his responses on a test measuring basic skills like how to count money, groom oneself, or use a microwave. He later argued that Plata “was just never taught and required to do a lot of these things. So I don’t think that’s a genuine deficit.”

Denkowski has argued that the test manuals produce skewed results because they offer no guidance on how to take such psychological and cultural factors into consideration. So he made up his own method and defended it in a 2009 article for the American Journal of Forensic Psychology, a journal that is not peer-reviewed. He reasoned that criminal offenders from poor families would not have learned such skills as counting money. Instead, they often develop skills that are more appropriate for a criminal lifestyle. And since the IQ and skills tests were “not developed for criminal offenders,” Denkowski argued that they fail to truly measure how well defendants have adapted. That makes it appropriate, in his view, to adjust test scores using one’s “clinical judgment.”

Denkowski made liberal use of his “clinical judgment” when he reported Plata’s scores. Judges rely on both adaptive-behavior and IQ scores. They are looking to see if both scores are above the cutoff point of 70. Denkowski adjusted Plata’s adaptive-behavior score from 61 to 71. He increased Plata’s IQ score from 70 to 77. He discounted testimony from Daniel’s parents and teachers, relying instead on information the prosecution gave him, including a list of books and magazines found in his cell and testimony from prison guards who said they saw him reading. (The guards later admitted they never heard him read aloud and had just seen him flipping the pages.)

Denkowski’s conclusion: Plata was “not considered to be mentally retarded for Atkinspurposes.”

When Kathryn Kase realized the prosecution was going to argue that her bumbling client could function like a normal adult — and could thus be executed — she knew she needed help. If she was going to win Daniel Plata’s Atkins claim, she had to find a psychologist who could prove that Dr. Denkowski’s methods amounted to junk science. So she contacted Jack Fletcher, a nationally known neuropsychologist who had served on the President’s Commission on Excellence in Special Education.

Fletcher watched the videotape of Plata’s testing and reviewed the test materials Denkowski had included with his opinion. His conclusion: Denkowski’s methods appeared to be “driven to yield scores outside the range of mental retardation.”

In December 2006, Plata’s hearing finally got under way before Federal District Court Judge Brock Kent Ellis in Houston. Both Denkowski and Llorente were quizzed about their opinions of Plata’s mental capabilities, and they critiqued each other’s methods. Ellis ended the hearing by saying that “both sides, all the lawyers involved, have worked extraordinarily hard in fully representing their clients in this regard and I appreciate it… I wish it was always the case.”

More than half a year passed before Ellis issued his opinion. He ruled that Plata is “a person with mild retardation” and should be removed from Death Row. More significantly, given Denkowski’s status as the “go-to” psychologist for Texas prosecutors, Ellis issued a scathing critique of his methods.

All of Denkowski’s testimony, the judge wrote, “must be disregarded due to fatal errors.” Ellis agreed with Fletcher that “it is not generally accepted practice within the field of psychological assessment to obtain an IQ score, declare it invalid, and then estimate an IQ score with numbers,” as Denkowski had done.

On Jan. 18, 2008, the Texas Court of Criminal Appeals agreed and commuted Plata’s death sentence to life in prison. He was transferred to the Hodge Unit in Rusk, where he is now housed with other mentally retarded prisoners.

The decision ended Daniel Plata’s 12-year stay on Texas’ Death Row. It might also lead to the end of George Denkowski’s career as a licensed psychologist. Judge Ellis’ decision emboldened one of Denkowski’s colleagues, Jerome Brown of Bellaire, to file a complaint with the Texas State Board of Examiners of Psychologists. If Denkowski loses his license, the cases of 17 other Texas men on Death Row — men he determined were not mentally retarded — could be re-examined. And Texas’ status as a national outlier in cases involving mental retardation could be changed for good.

Brown worked as an expert for the defense on five capital cases in which Denkowski worked for the prosecution. He says Denkowski used the same estimation techniques and showed the same deference to prosecutors’ evidence in those cases as he did in Plata’s, and that it was “essentially junk science. It is science that appears to be scientific, but it doesn’t have any background of validation to it.”

One of those cases resulted in an execution. In 2005, Brown and Denkowski tested Michael Richard, who had been sentenced to death for the 1986 rape and murder of a 53-year-old Houston woman named Marguerite Dixon. Based on test scores and school records, Brown concluded that Richard was mentally retarded, and had been all his life.

At first, Denkowski agreed that Richard was mentally retarded. As the state’s expert, he had submitted a finding that Richard had an IQ of 64 and adaptive-behavior scores that clearly showed mental retardation. His combined score was a 57, well below the 70 cutoff. But Denkowski retracted his findings after prosecutors showed him a list of books that were found in Richard’s cell, including two dictionaries. Denkowski said the dictionaries showed that Richard could read much better than he had indicated under testing. He adjusted several of Richard’s scores. When he added them up, the total score jumped from 57 to 76. In his new opinion, Denkowski concluded that Richard should no longer be considered mentally retarded.

When Brown saw the prosecution’s list of books, he met with Richard a second time to ask him about his reading abilities and clarify how he’d used the books in his cell — one of which was written in German. Denkowski had not followed up with Richard to ask about the books. Richard described to Brown how he stacked the books on top of each other and used them to sit on, since his death row cell lacked a chair.

Even so, the judge accepted Denkowski’s revised score. In September 2007, Richard was executed. Brown was appalled. “To those of us familiar with the right way to do these things, it is very apparent that what he’s doing is wrong.”

After the judge rejected Denkowski’s findings in the Plata case, Brown enlisted Jack Fletcher and a Florida-based psychologist named Tom Oakland to jointly file a complaint against Denkowski. Oakland co-authored the adaptive-behavior test. Their complaint cited the Richard case, as well as those of Plata and DeWayne Brown.

Last February, the state Board of Examiners of Psychologists upheld the complaint, finding that Denkowski had made “administration, scoring and mathematical errors” in all three cases. The board sent the complaint to the State Office of Administrative Hearings. Denkowski will have a chance to defend himself in a hearing scheduled for Feb. 16 in Austin. He could lose his license.

The broader psychological community has also rejected Denkowski’s methods. He is mentioned by name in the 2010 edition of the American Association on Intellectual and Developmental Disabilities’ diagnostic manual. In a section about how cultural or economic factors should impact scores on adaptive behavior tests, the authors “strongly caution against practices such as those recommended by Denkowski.”

Denkowski’s career as a prosecution expert appears to be over, whatever the outcome of his hearing. The Harris County District Attorney’s office stopped using him after Plata’s death sentence was overturned.

No other counties have hired Denkowski to work on Atkins cases since 2008. But the impact of his previous work continues to unfold. In October 2009, the U.S. Fifth Circuit Court of Appeals put the Atkins appeal of Steven Butler on hold pending the outcome of Denkowski’s hearing. In November a federal district court put another appeal, that of Joel Escobedo, on hold for similar reasons. And in December, the Fifth Circuit found that a lower court had “erred” when it dismissed Anthony Pierce’s right to a hearing based on Denkowski’s role in denying his Atkins claim. Several similar requests to put appeals on hold are pending.

Daniel Plata’s lawyer, Kathryn Kase, argues that all of the appeals on which Denkowski worked should be re-heard. “When you have junk science in a case, it’s like pouring poison into a punch bowl,” she says. “You aren’t going to get the poison out. So you have to pour out the punch, clean the bowl, and start all over again.”

Research support for this article was provided by the The Nation Institutes Investigative Fund. Special thanks to Gislaine Williams, Bryan Parras, Claire Loe and Tish Stringer for their assistance with translation and equipment.