Fallible evidence

“I don’t think a lot [of forensic science] has been tested in the courts yet,” says Gruspier from her office at the Ontario Forensic Pathology Service laboratories in suburban Toronto. Much of it, such as pattern evidence, does not have a grounding in hard science, she notes, but has been developed by the criminal justice system for its own use. And, “the trier of fact is deciding on evidence but is not a scientist.”

The test hinges in the reliability of the technique used — in other words, its fallibility.

Stephen Goudge, formerly an Ontario appellate judge, was commissioner for the Inquiry into Pediatric Forensic Pathology in Ontario in 2007-2008. The recommendations that he made in his final report included better training and qualifications for practitioners in the field of pediatric forensic pathology and paved the way for the creation of the Ontario Forensic Pathology Service, which now heads up all forensic pathology investigations in the province.

“The bigger issue [over new forensic techniques] is how forensic evidence is vetted by the justice system and how it’s evaluated,” says Goudge. “Is it reliable enough to act on?” There are groundbreaking types of science, he says, but the justice system needs to continue to evaluate it, and lawyers must remain skeptical of presented forensic evidence.

The DNA revolution

A few decades ago, it would have been the stuff of science fiction. Today, familial DNA searching, ancestry and phenotypic trait prediction from DNA, analysis of complex DNA mixtures and digital facial recognition are all being employed in crime labs.

In April, Joseph James DeAngelo was arrested in California on suspicion of being the Golden State Killer, who is believed to have committed a string of rapes and murders in California between 1974 and 1986. Police made the arrest based on DNA found at crime scenes that partially matched the DNA of a DeAngelo relation, found on the open-source genealogy website GEDmatch.

Familial DNA searching is “a cool new technique coming out of the analysis of complex DNA mixtures,” says Gruspier, adding that she isn’t yet sure how much impact it will have in the criminal justice system. Whereas DNA matching could once be performed only with close kin such as parents, children and siblings, today, science can match DNA from relatives further and further back, she says.

“Now, because the entire human genome has been mapped, you can go further away. . . . We’re also using new kinds of DNA [techniques] to predict what people look like and where they came from,” Gruspier adds, noting that a company in the United States can identify an individual from DNA, provide information about them and even generate a face. Ancestry and phenotypic trait predictions from DNA can be useful in identifying unidentified bodies, she notes.

And the analysis of complex DNA mixtures — from multiple DNA donors at a crime scene — now uses artificial intelligence to separate and identify the DNA. This is a newer technique that Gruspier calls “fascinating,” but she wonders how the courts will grapple with accepting the evidence of a machine over that of a human expert.

Probabilistic genotyping software is on the rise now, says Caitlin Pakosh, assistant Crown attorney in Hamilton, Ont. She is a former  lawyer at Innocence Canada and author of The Lawyer’s Guide to the Forensic Sciences. Also used for DNA evidence, probabilistic genotyping software’s purpose is to aid and interpret DNA profiles, including some DNA samples that might have been previously uninterpretable, she says.

“Ultimately, what it’s aimed at doing is . . .
[improving the] ability to improve samples, particularly complex samples” that involve DNA from more than one source, says Pakosh. The software uses math to evaluate the relationship between two different explanations of the DNA profile, she says. In August 2016, Ontario’s Centre of Forensic Sciences approved the use of STRmix, one such type of software.

Digital evidence

Digital facial recognition, facilitated by the omnipresence of surveillance cameras in today’s cities, is one type of digital forensic evidence now being employed by police services, which are using facial recognition software to identify perpetrators. Driver’s licences, health cards and other government-issued ID include photos of the bearers that are stored in government databases. A  computer can then make a comparison between a database photo and an image caught by surveillance footage.

In the past, Gruspier notes, courts have disallowed expert evidence on facial recognition performed by computers, but its popularity in police crime labs is changing that. However, she says, “No one’s tested it to see how often it’s accurate.”

Much forensic science has little scientific foundation, Gruspier adds. And while it’s been argued that cognitive bias can be eliminated by using computers, “somebody still has to program all that and decide what should be compared. I’m just naturally a bit skeptical.”

Emerging issues

Gruspier has similar concerns regarding fallibility with the so-called portable police lab and black box technology that are starting to be used at crime scenes. While it’s possible to run DNA through a black box — which will conduct a test and allow police to make an arrest on the spot — it is still a technology and, therefore, fallible, and even the DNA can be wrong, she says.

“I’m the DNA co-ordinator for a death investigation system,” Gruspier says. “I have to look at a probability number I get [from testing teeth, for example] to determine the match. If police are just throwing something into a black box . . . that whole interpretation part is gone.”

Pattern evidence — any markings produced when one object contacts another, such as fingerprints, bite marks, shoe prints and tool marks — is also problematic, Gruspier says, noting that individuals are on death row and have even been wrongly executed in the U.S. based on bite marks, to which a dentist will testify.

“We’re still struggling with the foundations of forensic science that have been accepted in our courts,” she says.

Privacy is also an issue as it concerns DNA evidence. The DNA Identification Act doesn’t allow for familial searching in the national DNA databank, says Pakosh (who stressed that views expressed were her own and not those of the Ontario Ministry of the Attorney General). “All of a sudden, now, there are public genealogy databases,” she says, such as the one that led to the arrest of DeAngelo.

While there are benefits to familial DNA searching — it can help identify suspects and solve cold cases, especially involving serial offences — the potential use of these public genealogy databases as an investigatory technique would mean that DNA samples would be collected in a manner currently outside the scope of the DNA Identification Act and the Criminal Code.

Before the courts

Courts have long sought certainty, and “in an effort to increase the level of certainty, we’ve often turned to forensic science to guide us,” says Jonathan Shime, a criminal defence lawyer at Cooper Sandler Shime and Bergman LLP in Toronto.

“It’s a noble effort, but the end result has been mixed, to say the least,” says Shime, who served as a staff lawyer to Goudge on the Inquiry into Pediatric Forensic Pathology in Ontario and whose partner Mark Sandler was the commissioned counsel. Although he agrees that there have been developments in forensic science from which the justice system has benefited, with DNA evidence being the most obvious, “The flip side is we have too easily been taken in, believing that forensic science . . . allows for greater certainty than reality would suggest,” he says.

Shime notes that high-profile cases of wrongful convictions in Canada that have been predicated on forensic science, such as the stomach contents evidence used to convict a young Steven Truscott of rape and murder in 1959 and the hair analysis used as evidence in Guy Paul Morin’s conviction for murder in 1992. (Improvements in DNA testing would exonerate Morin three years later.)

“The lesson [for] all of us in the justice system, whether you’re a defence lawyer, a Crown or a judge, is we need to be more diligent in how we think about what is ‘forensic evidence,’” says Shime.

When Gruspier testifies in criminal proceedings, she says, she is often asked about her bias. That’s important; but the question she isn’t asked is, “What is fallible in your evidence?”

And that may be the most important question to ask forensic witnesses. Lawyers should always meet with experts before they get to court, notes Shime. “If you only have one minute to meet with them, [the question] should be ‘What are the limitations of your opinion?’”

In short, the justice system’s challenge “is going to be to learn how best to evaluate what is offered as expert evidence,” says Goudge. A decade after his public inquiry concluded, there were serious problems with the way suspicious deaths involving children were handled in Ontario. “Is it reliable or not?”

Editor's Note: This article has been corrected where quotes and comments about portable police lab, black box technology and the scientific foundations of forensic science made by Kathy Gruspier were mistakenly attributed to Caitlin Pakosh. The errors also appeared in print in our October 2018 issue’s Forensic and Criminal Law report, “Fallible evidence.” Canadian Lawyer apologizes for the errors.

Forensic science today ideas

• Admissibility of evidence in trials

• Critique of pattern evidence

• Cognitive bias (by police, expert witnesses, judiciary)

• Scientific foundations of forensic science

Techniques

• Familial DNA searching

• Ancestry and phenotypic trait prediction from DNA

• Analysis of complex DNA mixtures

• Digital facial recognition

Emerging issues

• Portable police lab and black box technology at the scene

• Technological advances continue, but fundamental research is still lacking

• Digital forensic evidence

Provided by Dr. Michael Pollanen, chief forensic pathologist of Ontario and director of the Centre for Forensic Science and Medicine at the University of Toronto.