UPDATES ON CLPEX.com

No major updates on the website this week.

_________________________________________

we looked at a survey involving the discipline of forensic individualization based on images of detail other than friction ridge impressions.

we take a look at a recent article that is critical of latent print examination, as well as two replies to the article.

_________________________________________
Scholars Challenge the Infallibility of Fingerprints
by Peter Monaghan
http://chronicle.com
(http://chronicle.com)
Section: Research & Publishing
Volume 53, Issue 13, Page A14
November 17, 2006

Researchers find fault with a key tool in criminal investigations and find themselves ignored

For over a century, the recovery and analysis of fingerprints has been a key tool in criminal investigations.

The link between fingerprint and identity was forged in 1888 by Sir Francis Galton, a British scientist and mathematician who invented the science of fingerprint identification. Galton calculated the likelihood of two identical fingerprints at one in 64 billion, and ushered in the modern era of a practice that dated back to 14th-century China by noting that prints could be matched through the intersections, splits, and other "minutiae" formed by the ridges on the fingertips.

Today examiners still work with these minutiae, which are also called "ridge characteristics," "points of similarity," and "Galton details."

Galton assumed that each person's fingerprints were unique. But scholars such as Simon A. Cole, an assistant professor of criminology, law, and society at the University of California at Irvine, note that Galton's assertion has never been scientifically validated.

In numerous scholarly articles and a book, Suspect Identities: A History of Fingerprinting and Criminal Identification (Harvard University Press), Mr. Cole has argued that the problems with assuming that fingerprints are unique are compounded when forensic investigators and law-enforcement officers try to match prints that are often smudged or partial.

Researchers in other fields, including computer science and cognitive neuroscience, are also taking aim at what they see as the myth of infallibility that surrounds fingerprint identification. In particular, critics wonder: Why are the champions of the method so hesitant to submit it to scientific testing?

The lack of scientific scrutiny, says Mr. Cole, has broad implications for justice systems around the world. Because courts have largely sided with practitioners of fingerprint identification and not required more-scientific studies of the method, he observes, law-enforcement agencies "retain legal carte blanche to claim that fingerprinting is validated and infallible. They have nothing to gain and everything to lose from validation studies."

Mistaken Identities

Just what can go wrong when law enforcement depends on fingerprints was demonstrated in a high-profile terrorism investigation in 2004.

After the bombing of several passenger trains in Madrid on March 11, 2004, which resulted in 191 deaths and thousands of injuries, the Federal Bureau of Investigation arrested and detained a Portland, Ore., lawyer, Brandon Mayfield, on the basis of a "100 percent match" of his fingerprints with a set of fingerprints found on a plastic bag left near one bombing scene. Mr. Mayfield, a convert to Islam, was held in solitary confinement and then a jail mental ward.

But two weeks after his arrest, the FBI freed Mr. Mayfield after Spanish police correctly connected those same prints with an Algerian man.

Mr. Cole is spearheading efforts to test and improve the accuracy of fingerprinting. For instance, he points out that examiners rarely deal with whole fingerprints. They use "latent" prints: invisible impressions that a chemical agent converts into images, albeit often fragmentary, blurred, overlapping, or distorted ones.

Examiners then seek to match the minutiae in those prints to minutiae in much clearer inked or scanned prints in police databases that may hold millions of records.

Mr. Cole contends that such experts are far too credulous about finding or excluding matches. He says that the points upon which examiners base identifications are often poor because minutiae are obscured. Prints created by the same finger may look different, and prints from different fingers may look the same, creating errors at top laboratories and in systems based on complex computer-driven algorithms.

Researchers in other fields share Mr. Cole's skepticism about fingerprints.

For Anil K. Jain, a professor of computer science and engineering at Michigan State University, fingerprint identification is a pattern-matching problem that cannot be solved with any finality, but only improved upon. To find such improvements, Mr. Jain created a biennial contest — the International Fingerprint Verification Competition — in which he invites colleagues from academe and industry to present algorithms that improve the accuracy of fingerprint readings.

"When we started this competition," Mr. Jain says, "most of the time vendors would report their results based on their own proprietary data, and they could always clean up their reports." The current competition ends in January 2007, and Mr. Jain notes that the entries often give him and his colleagues a sense of how far they have to go in their own statistical analysis of how likely algorithms are to produce false matches when they analyze a specified number of "points" of identification on a finger. (The Federal Bureau of Investigation standard, for instance, is to use 12 of the 52 points of the human finger.)

Errors of Nature

As Mr. Jain evaluates the study of fingerprints with the aid of algorithms, Itiel Dror, a senior lecturer in cognitive neuroscience at the University of Southampton, in England, has looked at the human factor. Do the limits of human cognition create errors by fingerprint examiners?

Mr. Dror suggests that making such errors may be human nature. In experiments, he presented expert latent-fingerprint examiners with sets of prints about which they had previously made determinations. But he presented those prints to the subjects with new contexts, such as images of graphic violence.

Two-thirds of the experts "made differing, indeed conflicting, decisions," he reported in the journal Applied Cognitive Psychology last year and in the Journal of Forensic Identification this year. The context experiment "illustrates the vulnerability of humans, even highly trained and experienced experts, to cognitive and psychological effects that can influence and even distort their perceptions and judgments," he concluded.

The results are not surprising, he wrote: "The mind is not a camera; we are not passive in interpreting data."

Predictably, at his many presentations to gatherings of fingerprint examiners and law-enforcement officials, "some people get defensive," he says. "I can't blame them. I'm coming to people who have been doing it for many years and saying there is a problem. But I say to them, Yes, but people may have been wrongly sitting in jail for many years."

His message to such gatherings, he says, is that even when experts do their job optimally, mistakes happen because of normal cognitive processes. He also tries to elicit a promise from his audiences that they will use what information they have to improve the training of examiners.

"That," says Mr. Dror, "is a first, necessary step, but it's not enough." Also needed, he says, is the political and financial will of politicians and law-enforcement officials to support research on fingerprints and ensure that the findings make their way into training programs.

Eyeball Experience

Many critics do not imagine that research like Mr. Dror's will quickly shift opinion among fingerprint examiners. As academic research has undermined fingerprinting, scholars find that practitioners are tuning out. "It's not like a scholarly debate, scholar against scholar, where they fight it out in the literature," he says. "Scholars are used to scholarly debate, but in this dispute, the practitioners simply don't respond."

While it is not the job of practitioners to write for scholarly publications, Mr. Cole argues that fingerprint experts think the research "doesn't matter because it doesn't hurt them. They operate in the courtroom, where the scholarly literature is just ignored."

But fingerprint examiners will have to take at least some of their challengers seriously, says Glenn Langenburg, a forensic scientist with the Minnesota Department of Public Safety. He may be more open to academic research than his colleagues: He is also a doctoral candidate at the University of Lausanne, in Switzerland, where he is studying with Christophe Champod, a prominent researcher in efforts to develop statistical "probabilistic" models for the accuracy of forensic methods.

Signs that Mr. Langenburg is, as he says, "a minority in our profession" are easy to find. For instance, "a forum convened by the International Association for Identification in order to develop a fingerprint research agenda began its report by implicitly dismissing the need for validity study," he wrote in a recent essay, "stating 'This forum recognizes the reliability of friction ridge identification as practiced during the last 100 years.'"

Indeed, not all of the psychological research deflates examiners' claims to expertise. Thomas A. Busey, an associate professor of psychology at Indiana University at Bloomington, has demonstrated that examiners' repeated inspection of fingerprints does endow them with a visual expertise that improves their performance, compared with that of nonexperts.

While experts may have an edge, they sometimes believe they cannot make mistakes, several observers say. "Unfortunately, forensic scientists often reject error-rate estimates in favor of arguments that there is an error-free science," wrote two legal scholars, Michael J. Saks and Jonathan J. Koehler, in a Science review article last year.

"Likewise, fingerprint experts commonly claim that all fingerprint experts would reach the same conclusions about every print," wrote Mr. Saks, a professor of law and psychology at Arizona State University at Tempe, and Mr. Koehler, a professor in the business school at the University of Texas at Austin who specializes in behavioral decision making.

That claim, they say, amounts to saying that errors like the one that put Brandon Mayfield in federal custody are human errors, not methodological errors. And such claims are problematic, they contend, since the method relies "primarily on the judgment of the examiner."

And, they add, errors do, indisputably, occur, regardless of examiners' claims. In one set of industry tests, they note, about one-fourth of examiners made some errors, a result that even the editor of Journal of Forensic Identification, a leading industry publication, called "unacceptable."

Courting Controversy

The pressure for reform seems to be mounting, however. In the December 5, 2003, issue of Science, Donald Kennedy, editor in chief, echoed many of the objections that Mr. Cole and others have raised about fingerprints. He also noted that exaggerations of reliability may mar other techniques, such as hair sampling and analysis of bullet markings.

"Criminal-justice agencies," said Mr. Kennedy, "have been slow to adopt new scientific procedures and defensive about evaluation of their present ones."

The lessons learned by rigorous scientific examination of DNA analysis and polygraph testing were both salutary, he said. The former was proved reliable and accepted. But in 2003, the National Research Council of the National Academies found polygraph testing defective when they tested it on behalf of the U.S. Department of Energy, and recommended it not be used. (Nonetheless, he notes, the department rejected that recommendation and continued to use lie detectors.)

Still, the U.S. legal system has fought off challenges to fingerprint analysis. In 1993 the U.S. Supreme Court ruled that scientific testimony must be based on tested, dependable science. Many saw the ruling as a serious challenge to fields like fingerprinting, but two years later, a federal appeals court ruled that the new test did not apply to nonsciences — in that case, handwriting analysis.

A 1999 ruling by the Supreme Court overruled the federal appeals court and held that all expert testimony must meet tests of validity.

Fingerprint examination seems to have survived that ruling: In a decision last December, the Supreme Judicial Court of Massachusetts ruled that the test of validity for fingerprint analysis was not that the larger scientific community approved it, but that its own practitioners did.

The Way Forward

Scientific researchers such as Mr. Cole say they will continue to fight by filing amicus briefs in cases that adjudicate the admissibility of fingerprinting and other controversial evidence.

Mr. Saks and Mr. Koehler believe it is only a matter of time until fingerprint technicians adopt a more defensible stance toward the research that challenges their practice. They might follow the example of DNA typing, which involves the "application of knowledge derived from core scientific disciplines," which survived court scrutiny. For fingerprint science to achieve the same validity, critics say, it must make more use of such fields as differential geometry and topology.

Still Mr. Cole sees only a glimmer of progress in that direction. He observes that an FBI forensic-science report issued this year calls for the development of a research agenda. But, he says, "it's a carefully worded document; it doesn't give any ground on crucial legal issues."

***********************

RE: Scholars Challenge
by Michele Triplett

I find it humorous that an organization called “The Chronicle of Higher Education” would publish an article where the author didn’t do an adequate amount of research into his own topic.

-Galton invented the science of fingerprint identification?

-The FBI standard is to use 12 of the 52 points in the human finger?

-The 5 experts in Mr. Dror’s research (isn’t it Dr. Dror, Dr. Cole, and Dr. Champod?) were presented with images they previously ID’d under a new context, graphic violence?  I thought the new context was that the images were from the Mayfield erroneous ID?

These errors are actually irrelevant to the topic of the story, so I’ll try to ignore them and deal with the topic itself, that researchers feel like they’re being ignored.  The author points out that Cole says that uniqueness hasn’t ever been scientifically validated.  It looks to me like Monaghan (the author) never looked for any information to the contrary.  If he had looked, he would have quickly found that multiple researchers have studied 1) the embryonic development of the skin (I’m not listing them because this information is easily available); 2) several statistical models exist supporting uniqueness (I know, these models don’t accurately represent reality - that’s why we call them models, they resemble reality, they don’t exactly replicate it).  And 3) empirical data collected for over 100 years also scientifically supports the theory that fingerprints are unique.  If anyone thinks that the latent print community has been silent about this, they’re simply not listening.  We’ve said it and we’ve screamed it.  Most of us have finally realized that screaming any louder won’t help (ok, so I haven’t really realized it yet J). 

Why are we so hesitant to submit to scientific testing?  First, research is being done.  Scientific research requires time and money, it hasn’t been all that long since requests for additional research were asked for.  The latent print community has acknowledged this need and has responded (although not as strongly or as quickly as many would like) .  Good conclusions don’t happen overnight, high quality scientific research sometimes takes decades.  With regard to uniqueness and additional research, as a comparison for Mr. Monaghan, just because a few quacks insist the earth is flat, does that mean that the scientists have to continue spending their time and research money trying to convince them? 

Infallibility, absolute conclusions, and 100% match……these statements were common in the past but our industry has recognized that perhaps the industry was overstating results.  We have acknowledged that we are human and errors are possible.  We are continually looking at ways to diminish errors but this doesn’t mean that the error rate is high.  Human error is always going to exist but if you look at the data, the error rate of this industry is amazingly low!

One more comment.  Since I’m criticizing the author and the critics, in all fairness I should also make a few comments about one of our own (Glenn Langenburg).  The article states that Glenn Langenburg may be more open to academic research than his colleagues.  I doubt if that’s true.  We’re all open to bettering the industry and continually making improvements, Glenn’s just more motivated than the rest of us (a lot more motivated!!).  The author makes it sound like Glenn is the only one who recognizes the need for validity studies.  This is supported by stating that an IAI report about creating a research agenda stated that reliability is already recognized (I paraphrased this).  Anyway, I think we all recognize the value of validity studies but at this time, the need for validity studies is probably higher in other areas.  If time and money are going to be appropriated, it only makes sense that it gets appropriated in the most needed areas first.

I don’t blame the critics for misrepresenting the facts in this article.  I would image that they are rolling their eyes in their head as well.  I blame the journalist.  This article is nothing more than a regurgitation of previous statements and questions that have been asked and answered.  I also feel like it’s filled with glaring errors that make most of us discount the article as a whole.

Michele Triplett

***********************************

RE: RE: Scholars Challenge
by Glenn Langenburg

Michelle afforded me an opportunity to read her critique of the article.  I agree whole-heartedly with Michele's observations of the misrepresentation of the author.  I think it is unfair to paint me as one of the only people in the profession open to research.  If I recall, from the interview, we were talking about the use of a probabilistic approach and whether or not the error rate is "zero".  I am hardly the only one in the profession that is interested in addressing these issues.  If that were true, this would truly be a sad state of affairs for our discipline.  There are many notable individuals driving this profession forward with their research.  It takes time.  Science cannot be rushed to fit the courts', critics', or other nay-sayers' time tables.

My only comment is, if some of my statements were taken out of context, and misrepresented, then I take any of the quotes from critics in the article with a grain of salt as well.   Thank-you Michele for your critique.

Glenn Langenburg