Shaken Baby Syndrome: How Bad Science Can Result in False Confessions

For the past decade, the theory of shaken baby syndrome has been under attack. See here and here.  In January, the U.S. District Court of the Northern District of Illinois granted Defendant Jennifer Del Prete’s habeas corpus petition where her conviction for first degree murder was based on faulty evidence of shaken baby syndrome.  Del Prete v. Thompson, 2014 WL 296094 (N.D. Ill. Jan. 27, 2014).  In a footnote to the opinion, the court even wrote that the testimony and evidence presented “arguably suggests that a claim of shaken baby syndrome is more an article of faith than a proposition of science.”

Incriminating statements in Del Prete illustrate how the power of supposed medical evidence can lead to false confessions.  The defendant was a child care worker.  She explained to law enforcement that the baby was not responding, and she picked her up and gave her a “very slight shake.”  Still getting no response, the defendant gave the child pats on the back to attempt to dislodge something possibly choking her and later administered CPR.  The detective employed confrontational interviewing techniques, brought up the concept of shaken baby syndrome, and told the defendant that her statements were not consistent with what the police had learned from doctors.  Upon hearing that her story did not match the medical evidence, the defendant cried and then said that she “couldn’t remember ‘exactly how things took place’ because of the increasing panic she was experiencing.”  Unable to provide an explanation, the defendant said, “[E]ven if I was panicked, aren’t I responsible? Am I going to go to jail?”  While the defendant did not make a confession, the pressure from the interrogation caused her to make assumptions about her own responsibility or culpability and made her question her own memory.

In a similar case, Judge Posner of the Seventh Circuit found that a defendant’s confession to shaking was “worthless as evidence.”  Posner found, “Not being a medical expert, Aleman could not contradict what was represented to him as settled medical opinion. He had shaken Joshua, albeit gently; but if medical opinion excluded any other possible cause of the child’s death, then, gentle as the shaking was, and innocently intended, it must have been the cause of death.  Aleman had no rational basis, given his ignorance of medical science, to deny that he had to have been the cause.”  Aleman v. Vill. of Hanover Park, 662 F.3d 897 (7th Cir. 2011).

The hypothesis of shaken baby syndrome is especially susceptible to false confessions because the hypothesis itself can only be validated or confirmed by a confession.  Interrogations often involve parents and caregivers in situations of extreme trauma and vulnerability, having experienced the death of a child only hours beforehand.  When faced with what is presented as absolute medical evidence that shaking is the only possible explanation, parents and caregivers give coerced compliant false confessions (“If you say so, I must have done it.”) and internalized false confessions (where a suspect becomes convinced that he or she has done it).  This risk of false confessions creates a circuitous problem—the presentation of unreliable medical evidence may cause a suspect to make false or overstated admissions, but even if this unreliable medical evidence is challenged at trial, the prosecution may still be able to rely on the admissions of the defendant to obtain a conviction.

Further resources:

• Tuerkheimer, Deborah, Flawed Convictions: “Shaken Baby Syndrome” and the Inertia of Injustice (Oxford University Press 2014).
• Cavazos v. Smith, 132 S.Ct. 2 (2011) (dissent states that “doubt has increased in the medical community ‘over whether infants can be fatally injured through shaking alone.’”).
•  Tuerkheimer, Deborah, The Next Innocence Project: Shaken Baby Syndrome and the Criminal Courts, 87 Wash. U. L. Rev. 1 (2009).
•  Keith Findley et al., Shaken Baby Syndrome, Abusive Head Trauma, and Actual Innocence: Getting it Right, 12 Hous. J. Health L. & Pol’y 209 (2012).
•  Shapiro, Joseph, Rethinking Shaken Baby Syndrome, June 29, 2012.
• Brief of State of Wisconsin v. Quentin Louis

Cognitive Bias and Forensic Anthropology

A study looking at how the conclusions of forensic anthropologists may be influenced by extraneous information highlights the importance of protecting all scientists from potentially biasing information.

Forensic anthropologists determine the gender, national origin, and age of a person at the time of death. In some cases this determination must be based solely on skeletal remains. A recent study introduced extraneous information to the scientists when remains were submitted for scientific analysis and found that background information can bias a scientist’s conclusions and result in mis-identifications across the board.

In this study, three test groups were assembled to analyze the remains of a nineteenth century female skeleton with strikingly neutral, or “ambiguous,” features which could allow a conclusion of either sex. An ambiguous environment is one that creates the greatest potential for cognitive bias. Cognitive bias occurs when contextual information influences a person’s thought process and may lead to inaccurate judgment. In this case, scientists relied on information that was submitted with the evidence, causing them to make conclusions validating or complementing the extraneous information.

Two of the three groups were given DNA profiles in addition to skeletal remains. The first group’s DNA profile indicated the skeletal remains belonged to a Caucasian male between 25-30 years of age. The second group was given a DNA profile indicating the same skeleton was an Asian female between 50-55 years of age. The third group, the control group, was given no supplementary information. Each scientist, ranging from master’s degree students to practicing PhDs in anthropology, was given up to an hour to examine the remains and make conclusions. In the first group, 72% concluded the remains were male and 100% of examiners in the second group concluded the remains were female. In the control group 31% of examiners concluded the remains were male and 69% concluded female.

Regarding ancestry, 100% of group one (who were told the skeleton was Caucasian) and 100% of the control group (who were given no extra information) decided that the skeleton was Caucasian. In group two, where the scientists were told the skeleton was of Asian ancestry, 50% concluded the skeleton was of Asian descent and 50% concluded Caucasian. The study attributes the bias in both group one and the control group to cognitive bias and bias generated from ancestrally-common characteristics described in anthropological textbooks.

This study showed that age at time of death is the least affected category by cognitive bias. Anthropologists in each group reached conclusions about age independent of the supplied information.

This study shows how biasing contextual information can affect a scientist’s ability to be impartial. The accuracy and credibility of a scientist’s conclusions depends upon blinding the scientist to any information that can be consciously or unconsciously biasing. Attorneys may be interested in reading similar studies that have looked at how contextual information can bias examiners in the fields of blood spatter interpretation, fingerprint comparison, and DNA interpretation.